Skip to the Main Content

Note:These pages make extensive use of the latest XHTML and CSS Standards. They ought to look great in any standards-compliant modern browser. Unfortunately, they will probably look horrible in older browsers, like Netscape 4.x and IE 4.x. Moreover, many posts use MathML, which is, currently only supported in Mozilla. My best suggestion (and you will thank me when surfing an ever-increasing number of sites on the web which have been crafted to use the new standards) is to upgrade to the latest version of your browser. If that's not possible, consider moving to the Standards-compliant and open-source Mozilla browser.

September 2, 2006

TeXnical Issues

Posted by John Baez

Thanks to Jacques Distler, this blog uses some marvelous technology that lets you write comments with equations in them. It’s based on TeX. The TeX - or more precisely itex - is converted to MathML.

As with any sufficiently advanced technology, you can waste a lot of time trying to figure out why the hell it isn’t working.

Here you can read a little FAQ, followed by 400 questions and answers in the form of blog comments. I have disabled further comments because this particular entry is a favorite target of spammers. So, if your question is not answered in what you see below, please email me.

(Figuring out how to do this is a little intelligence test to weed out spammers.)

n-Category Café – TeXnical FAQ

Here are answers to some really basic problems you may have with this blog.
  • Why do all the equations look like gobbledygook?

    Maybe you don’t know math, so all equations look like gobbledygook.

    More likely, you forgot to download the necessary fonts. To do this, read Jacques Distler’s advice. He tells you what to do, depending on what operating system and browser you use. If you use Windows, I recommend switching to Firefox if you haven’t already; then installing the requisite fonts is easy. Your mileage may vary — if you get stuck, see the discussions below, especially this.

    You can check to see you have all the necessary fonts by going to Jason Blevin’s test page. Look at all the symbols, and see which look okay.

  • Why doesn’t it work when I make comments using TeX?

    When you post comments, you are given a choice of “Text Filters”. The default choice is “Convert Line Breaks”. That’s good if you don’t want TeX. But if you want TeX, it won’t work! For TeX, I suggest

    • “itex to MathML with parbreaks” if you like writing in XML, or

    • “markdown with itex to MathML” if you prefer a simple markdown system.

    If you’re used to TeX, please keep in mind that on this system, TeX kicks in only inside equation environments. Outside those, your chosen text filter is in charge!

    There are lots of other things that can go wrong, since itex and MathML are sort of complicated. But, if you keep clicking on “preview” and struggling to decipher the cryptic error messages, you’ll either figure out your problem… or you won’t.

  • Why doesn’t it work when I include HTML links?

    Maybe you’re using “A HREF” instead of the lower-case “a href”. Unlike HTML, the system here is case-sensitive. This should work:

    <a href = "http://math.ucr.edu/home/baez/TWF.html">This 
    Week's Finds</a>

    This would not:

    <A HREF = "http://math.ucr.edu/home/baez/TWF.html">This 
    Week's Finds</a>

    Since we’re engaged in scholarly discussion, I encourage including references with links. If you want to make me really happy, something like this is great:

    <ul>
    <li>
    Ross Street,
    <a href = "http://arxiv.org/abs/math.CT/0303175">Categorical and 
    combinatorial aspects of descent theory</a>.
    </li>
    </ul>
    

    It should give you something like this:

  • Why doesn’t it work when I use “blockquote” to quote someone?

    Again you need lowercase, but also, when using the text filters “Convert Line Breaks” and “itex to MathML with parbreaks”, you need to skip a line before and after the “blockquote” command. So, with these text filters, this should work:

    John Baez said:
    
    <blockquote>
    
    In the 60's, Grothendieck led the reggae group 
    shown in this rare photo:
    
    </blockquote>
    
    

    This would not:

    John Baez said:
    
    <blockquote>
    In the 60's, Grothendieck led the reggae group 
    shown in this rare photo:
    </blockquote>
    

    Other text filters have other demands.

    You may also run into trouble if you just cut and paste the text you’re trying to quote, because it may contain math or other special characters - see below.

  • How come weird stuff happens when I use symbols like & or < ?

    Some symbols have special meanings in HTML. To keep the computer from thinking you intend those special meanings, instead of typing a raw & you have to type

    &amp;

    which is the HTML code for an ampersand. Similarly, instead of < you need to type

    &lt;

    (Extra credit if you can figure out what I have to type for you to see "&amp;".)

  • Can I customize my view of the comments, to help keep track of the mammoth conversations on this fascinating blog?

    A little - at the bottom right of the list of comments, you can click on a little button saying “view chronologically” to see the comments in chronological order. Then it says “view threaded”; clicking on it again restores the original view.

  • Can I customize my view of the comments a lot more, get the comments form to automatically choose “itex to MathML with parbreaks” every time, and stuff like that?

    Only if you’re smarter than me. If you’re using Firefox, you can download Greasemonkey, an add-on which lets you customize your web-browsing experience using little programs called scripts. Mike Stay has written a Greasemonkey script specially for viewing this blog. You can get it here.

Posted at September 2, 2006 8:46 AM UTC

TrackBack URL for this Entry:   https://golem.ph.utexas.edu/cgi-bin/MT-3.0/dxy-tb.fcgi/916

400 Comments & 2 Trackbacks

blockquote

Concerning the blockquote-tag, the point is that it does require the content included in <blockquote>…</blockquote> to be inside a paragraph environment enclosed by <p>…</p>.

So you either have a text filter chosen which converts line breaks to such <p>…</p> tags and then insert line breaks before and after the blockquote-tag as John says above – or you add in the tags by hand and do

<blockquote><p>text goes here</p><blockquote>

And if the quoted text has itself several paragraphs you type


<blockquote>
<p>
first paragraph
</p>
<p>
second paragraph
</p>
<blockquote>

which produces this output

first paragraph

second paragraph

Posted by: urs on September 2, 2006 2:09 PM | Permalink

Re: blockquote

If you’re not keen to type XHTML markup, you can also choose the Markdown (without TeX equations) or Markdown with itex to MathML (with TeX) filter and type

> This is a blockquoted paragraph
>
> This is *another* paragraph.
>
>> This is a blockquote **within** a blockquote.
>
> And yet one more paragraph.

to obtain

This is a blockquoted paragraph

This is another paragraph.

This is a blockquote within a blockquote.

And yet one more paragraph.

Also, I’d like to point out that development of the itex2MML filter, that converts TeX to MathML here, is very much user-driven. I’ve tried to implement the most useful subset of LaTeX (with a few MathML-inspired extensions, like \tensor{R}{_\mu_^j^k_l} and \multiscripts{_2}{F}{_1}, which produce Rμ j k l \tensor{R}{_\mu_^j^k_l} and F1 2 \multiscripts{_2}{F}{_1}, respectively). But, if there are some TeX commands you wish were implemented, but aren’t, let me know.

As to fonts, the current situation is pretty ugly. My best understanding is here. When the Stix fonts are released, we will finally have a complete set of mathematical symbol fonts, which work on all platforms (Linux, Mac and Windows). But that’s still a ways in the future.

Posted by: Jacques Distler on September 2, 2006 7:10 PM | Permalink | PGP Sig

Making the font package

I’ve tried several times to get this to work, with no success. I’m running the xft-disabled build of firefox on Linux (I think). I try to follow the instructions for creating the font packages here.

I discover that there is no “type1inst” command on my distribution (Fedora Core 5). I find a “type1inst” here and install it manually. Following instructions by running type1inst on the cmpsfont/pfb, I get messages:

94 fonts found
94 were standard PostScript fonts
——————————————————-
For 94 of these I couldn’t figure out which foundry
the font is from. Thus, these fonts will appear under the
foundry unknown, i.e. X font name -unknown-*.
Please consult the README file to see what this means.

I don’t know if this is a problem or not.

When, as instructed, I run type1inst on the PCF/ directory, I get the message:

There are no PostScript fonts in this directory

So I’m not surprised when I follow the rest of the instructions, and fonts don’t work. (Add to that the fact that my up-to-date Linux distribution seems to handle fonts in a differnt way than the older ones described in these instructions … and, of course, I don’t know anything about fonts …)

Has anybody ever gotten this to work? I’m curious.

Posted by: Charles on September 2, 2006 5:02 PM | Permalink

Re: Making the font package

I’m not sure exactly what you are going for here, but if your goal is to get mathml to display correctly in Firefox, under Fedora Core 5, it’s much easier than what you are trying to do.

First of all, I assume you have root access on the machine you are working on. If not, find a system administrator and ask them nicely to help you. Now, check to see if you have the mathml fonts installed with the following command:

rpm -q mathml-fonts

If they are installed it will return the name of the package. If not, it will say “package mathml-fonts is not installed”. If it is not installed then you can grab it from:

mathml-fonts

This can be installed (as root) using the command:

rpm -ivh xxxx

where “xxxx” is the name of the mathml-fonts rpm you downloaded, above. Alternately, if you are comfortable using “yum” you can find it in the “extras” repository.

Installing the mathml-fonts package should provide the fonts you need. They are truetype fonts, so if you need type 1 fonts for something else then this will not help you.

There is one more thing you have to do. By default, Fedora builds firefox with “Pango” support for text layout. For various reasons this doesn’t quite jive with mathml rendering. Hopefully this will be fixed soon. As root, you need to open the file /usr/bin/firefox, which is actually a script that runs firefox with all the proper variables set, and comment out the lines that tell firefox to use Pango. I think they are somewhere around line 84 in the file. That section of the file will look like this:

#
MOZ_DISABLE_PANGO=1
export MOZ_DISABLE_PANGO
#

You should change it to look like this

#
# MOZ_DISABLE_PANGO=1
# export MOZ_DISABLE_PANGO
#

Once this is done, save the file.

Now when you run firefox it should properly display mathml.

Posted by: Bob McNees on September 3, 2006 6:48 PM | Permalink

Re: Making the font package

I think you meant it the other way around, i.e. you want to remove the comment marks on the lines that disable pango.

Posted by: Mike on September 4, 2006 2:10 AM | Permalink

Re: Making the font package

Whoops! That’s embarassing!

Yes, I switched things around. Initially, that section of the file /usr/bin/firefox will look like this:

#
# MOZ_DISABLE_PANGO=1
# export MOZ_DISABLE_PANGO
#

and it should be altered (this must be done as root!) to look like this:

#
MOZ_DISABLE_PANGO=1
export MOZ_DISABLE_PANGO
#

After this change has been made firefox will no longer use pango for text layout, and mathml should display correctly.


Posted by: Bob McNees on September 4, 2006 3:41 PM | Permalink

Re: TeXnical issues

You can’t type $x^{*}$ in markdown with itex to mml, because it thinks the asterisk is for emphasis. I got around it by using $x^{\star}$.

Posted by: Mike on September 2, 2006 7:29 PM | Permalink

Markdown Filter

Yikes!

If so, that’s a bug in the filter. Markdown is supposed to ignore things like “*” when they occur within MathML markup.

Thanks for bringing this to my attention. It must be some kind of regression, as this didn’t used to be a problem. I’ll check into it, and hopefully get it fixed ASAP.

Posted by: Jacques Distler on September 2, 2006 7:37 PM | Permalink | PGP Sig

Markdown filter

I’m not reproducing your problem, Mike. Here is x x^{*} ($x^{*}$) and here is the same thing as a display equation x x^{*} ($$x^{*}$$). Both seem to work just fine for me.

Posted by: Jacques Distler on September 2, 2006 7:56 PM | Permalink | PGP Sig

Re: Markdown filter

Try that expression all by itself. I get an error and find this in the offending line:

<p>$x^{<em\>}$<math xmlns='http://www.w3.org/1998/Math/MathML' display='inline'><msup><mi>x</mi> <mo></em></mo></msup></math></p>

Clearly the asterisk has been replaced by <em/>.

Posted by: Mike on September 3, 2006 4:01 AM | Permalink

Re: Markdown filter

There’s a bug in itex2MML that I haven’t been able to track down. If the very first character of the input is a “$” (because you happen to start your comment with a formula) then it echoes back the TeX formula, in addition to the processed MathML output.

This (for obvious reasons) screws up the Markdown filter. For silly reasons, the itex to MathML with parbreaks filter is not affected.

The workaround is trivial; Insert a blank space before the first $, and the Markdown filter will work correctly.

Posted by: Jacques Distler on September 3, 2006 4:21 AM | Permalink | PGP Sig

Re: TeXnical issues

Why doesn’t it work when I include HTML links?

If you’re using markdown, then you’d write links like this:

 A neat paper by <math xmlns='http://www.w3.org/1998/Math/MathML' display='inline'><semantics><mrow><mo stretchy="false">[</mo></mrow>[</semantics></math>Abramsky and Coecke<math xmlns='http://www.w3.org/1998/Math/MathML' display='inline'><semantics><mrow><mo stretchy="false">]</mo></mrow>]</semantics></math><math xmlns='http://www.w3.org/1998/Math/MathML' display='inline'><semantics><mrow><mi>http</mi><mo>:</mo><mo stretchy="false">/</mo><mo stretchy="false">/</mo><mi>arxiv</mi><mo>.</mo><mi>org</mi><mo stretchy="false">/</mo><mi>abs</mi><mo stretchy="false">/</mo><mi>quant</mi><mo>&minus;</mo><mi>ph</mi><mo stretchy="false">/</mo><mn>0402130</mn></mrow>http://arxiv.org/abs/quant-ph/0402130</semantics></math>

and they’d appear like this:

A neat paper by Abramsky and Coecke

You can put a backslash before most symbols to get them to appear, but using the markdown with itex to mml filter, \[[ means “center this equation” so I wrap the bracket in dollar signs. I don’t know how to escape a dollar sign under this filter.

Mike Stay has written a “greasemonkey script” that lets you do stuff like this if you’re using Firefox. Don’t ask me what that means.

Firefox is a web browser. Greasemonkey is a plugin for the browser that allows you to preprocess your web pages before seeing them. People write up little programs, called user scripts, that rewrite the web page in a way that’s more useful to them. For example, you can remove ads, rewrite links to go to “printer” versions of articles instead of the click-through ones, change color schemes that are hard on your eyes, etc.

My user script hides all the comments except the one you’re responding to, and also sets the filter to markdown with itex to mml.

Posted by: Mike on September 2, 2006 7:49 PM | Permalink

Choose wisely

You can put a backslash before most symbols to get them to appear, but using the markdown with itex to mml filter, \[ means “center this equation” so I wrap the bracket in dollar signs. I don’t know how to escape a dollar sign under this filter.

The easy answer is: if you’re not using math in your comment, choose the ordinary Markdown filter, where these characters don’t have any special meaning. Then you can just type “$” and “\\[”.

If you really need those characters (outside of equations) and you need equations in the same comment, you can always type “&#x24;” ($) or “&#x5c;[” (\[).

Posted by: Jacques Distler on September 2, 2006 8:10 PM | Permalink | PGP Sig

Re: TeXnical issues

Thanks, folks! As time passes I will try to update the FAQ to provide better answers to the questions, and answers to more questions - but probably always with an emphasis on the questions of the most clueless sort, since I’m sure there are lots of good mathematicians and physicists who are flummoxed by the simplest things here.

I’m afraid that “TeX” and “posting to a blog” are sufficiently scary to prevent a lot of good people from joining in the conversation, because I’ve been getting a certain number of comments via email. I don’t now have the energy to try to help people overcome those barriers, but maybe someday I can include links to places that explain that stuff.

I’m just curious, Jacques - which do you find to be the most efficient filter for posting to your blog? Or do you pick one based on what you’re trying to do? Right now I’m using “itex to MathML with parbreaks” except for posts with no math at all.

Posted by: John Baez on September 3, 2006 12:10 AM | Permalink

Choice of tools

For posting to my blog, I mostly use a blogging client, ecto.app, for MacOSX. It has many nifty features (like user-defined macros for inserting XHTML markup), which make composing posts much more pleasant.

The huge disadvantage is that it uses a non-MathML-aware renderer for its preview function. So I often end up correcting the equations in my posts, after publishing them.

As to filters, I tend to use whichever one seems most appropriate.

For comments, that’s usually either the default (Convert Breaks) or one of the Markdown filters (with or without itex to MathML)

For posts, since I have macro-keys to insert whatever markup I typically use, I tend to rely more on the itex to MathML with parbreaks or the Convert Breaks filters, depending on whether the post has math in it.

Just about the only filters I never use are the Textile filters. But some people like 'em, so I keep them around …

I wish I could recommend one single filter as all-around best, but I can’t.

As much as I like Markdown syntax, for instance, it doesn’t support the cite="..." attribute for <blockquote> elements. I like supplying the URL for something I quote. And you’ll notice that those cite attributes get handled in a particularly nifty way hereabouts.

Posted by: Jacques Distler on September 3, 2006 3:59 AM | Permalink | PGP Sig

Re: TeXnical issues

Long ago, Arthur C. Clarke famously stated that “any sufficiently advanced technology is indistinguishable from magic”, leading Gregory Benford (among others) to conclude that any technology distinguishable from magic is insufficiently advanced. By the same logic, I suppose we should adopt Baez’s Nth Law, “Any technology which does not break in fantastically incomprehensible ways is so insufficiently advanced it’s not worth bothering about.”

Posted by: Blake Stacey on September 3, 2006 12:16 AM | Permalink

Re: TeXnical issues

Nah. Technology that doesn’t break in fantastically incomprehensible ways is obviously magic, ergo the the previously described category of “insufficiently advanced” is sufficient.

But kudos to all of you for pushing mathml with all the attendent development work! I always thought that it was deeply ironic that for almost a decade after the web was invented for the distribution of scientific communication, we didn’t have a reasonable way to include mathematics. Nice to see this finally changing.

Posted by: Ralph Giles on September 4, 2006 1:11 AM | Permalink

Re: TeXnical issues

I wonder if in a few years time those involved in talking about maths on the internet will lament the passing of days when they had to use ASCII. Unlikely, I guess. More reasonable though to lament the passing of the blackboard. Sasha Borovik has a good post on this.

As I predicted, Borovik’s blog is proving very interesting.

Posted by: David Corfield on September 4, 2006 12:59 PM | Permalink

related issues

In one of the comment sections, John wrote

I think I’ll keep posting the “main line” of my Socratic dialog with Urs as comments that will appear on the very bottom of this page, instead of commenting on a comment on a comment….

That’s fine with me.

I just remark that this need not be an either-or decision. Using the button “view chronologically” found at the bottom of each comment section we can always turn the tree of comments into a linear list and see our latest comments appear at the bottom.

Therefore personally I tend to use the nested threading, becasuse - if desired - this is useful for disentangling parallel discussions in a long comment section.

Posted by: urs on September 6, 2006 2:41 PM | Permalink

Re: related issues

Good point. As you’ve probably noticed, I like to sort of “manage” the discussions about my blog entries by writing long comments that sum up what’s been said so far, explain stuff, and pose puzzles. I’d like to have these long comments stand out visibly, but have also trees of discussion consisting of shorter posts where people (including me) are busy talking to each other about millions of things. That’s what I’m trying to achieve. But maybe I’m just being too bossy, trying to run the show too much.

Posted by: John Baez on September 6, 2006 4:20 PM | Permalink

Re: related issues

But maybe I’m just being too bossy, trying to run the show too much.

No, that’s great.

I just thought I should also mention this “threaded/chronological” toggle. On some browsers it is sometimes hard to see (IE) - and it’s a cool feature.

Posted by: urs on September 6, 2006 4:47 PM | Permalink

Re: TeXnical Issues

Is it just me, or did you change the way preview works recently? It doesn’t have the parent_id as part of the URL and thus my greasemonkey script has no way to pick out the message. (Maybe I just never noticed that it only works on the first screen after hitting “Reply to this”.) Is it possible to make the parent_id part of the url when previewing comments?

Posted by: Mike on September 6, 2006 9:21 PM | Permalink

Re: TeXnical Issues

Is it possible to make the parent_id part of the url when previewing comments?

The arguments to an CGI script are appended to the URL, when accessed via an HTTP GET. They are part of the body of the request, when accessed via an HTTP POST.

For sound technical reasons, the comment form is (and always has been) submitted as an HTTP POST.

I don’t see why this should be an issue. There are plenty of other ways greasemonkey could keep track of this information, besides looking at the current URL.

Posted by: Jacques Distler on September 6, 2006 11:07 PM | Permalink | PGP Sig

Re: TeXnical Issues

I’ve changed my script so that it fetches it out of the hidden input tag when it’s not in the URL.

Posted by: Mike on September 7, 2006 6:47 PM | Permalink

Re: TeXnical Issues

Now it also inserts a little button to show / hide each comment; by default, it only shows the one you’re responding to, but you can click now to see others.

Posted by: Mike on September 8, 2006 12:49 AM | Permalink

Search

Is search broken for anyone else?

Posted by: Mike on September 6, 2006 11:00 PM | Permalink

Re: Search

Yes, it seems busted to me. I typed in “doctrine” and then “n-category”, and both times I got:

Movable Type

An error occurred

You are currently performing a search. Please wait until your search is completed.

with no sign that anything would ever happen.

Posted by: John Baez on September 7, 2006 4:32 PM | Permalink

Re: Search

Yet another thing that was broken by the upgrade to MT 3.3x.

For historical reasons, there’s a throttle on the number of simultaneous searches that can be performed.

This worked quite flawlessly through many revisions of MovableType but seems to have broken with MT 3.3x. I’ve gone in and restored the previous behaviour.

Let me know if you have any further problems.

Posted by: Jacques Distler on September 8, 2006 5:24 AM | Permalink | PGP Sig

Input box

Is it a design choice that the box where you type in comments is rather small, or is it just a facet of my particular configuration?

As I see it (in Firefox on Linux), it’s 42 characters wide and 10 high. To me this feels quite claustrophobic, and I’d prefer something much bigger. For instance, when I use a text editor for typing Latex etc, it’s 75 characters wide and 35 high, and I find that comfortable.

Thanks.

Posted by: Tom Leinster on October 14, 2006 2:53 PM | Permalink

Re: Input box

Good point. My box is the same size, and indeed this has always annoyed me.

I’ll ask Jacques how easy this is to change.

Posted by: John Baez on October 20, 2006 11:34 PM | Permalink

fonts… and css

Why do all the equations look like gobbledygook?

Well, what I found in firefox (and mozilla) under Solaris, is that it was gobbledygook even after installing the fonts. The problem is the line for the font-family in the file …/mozilla/res/mathml.css; it gives Symbol precedence over every other math type except CMSY10, and this one is specifyed in uppercase, btw. Putting Symbol at the end of the line seems to solve the gibberish look. Tip from Ian Hutchinson.

Posted by: Alejandro Rivero on October 22, 2006 4:40 PM | Permalink

Two Tips for Perfectionists

For the aesthetic totalitarians among you, who like everything as beautiful as humanly possible:

  1. Due to some slight bug, letters in math formulas come out in roman font if no space is left between them. Thus, if you multiply the variables xx and yy, you’ll get xyxy instead of xyx y unless you put in a space.
  2. Use only right-handed single quotes if you want

    ‘quotes like this’

    If you use a left-handed single quote on the left side, the way you learn to do in TeX, you’ll get

    `quotes like this’,

    which aren’t nearly as pretty.

I hope nobody fixes these ‘bugs’, now that I’ve written reams of material that’s adapted to them.

Posted by: John Baez on December 1, 2006 5:59 AM | Permalink

Re: Two Tips for Perfectionists

Due to some slight bug, letters in math formulas come out in roman font if no space is left between them.

“It’s not a bug; it’s a feature.”

The MathML Specification recommends that the multiletter token, <mi>xy</mi> be rendered in an upright font, whereas the single-letter tokens <mi>x</mi><mi>y</mi> should be rendered in italics.

Question: how to access this feature from itex? TeX doesn’t normally have a concept of multi-letter tokens: $xy$ is treated as two tokens, just like $x y$. In this regard, itex follows the MathML convention. $xy$ is a single two-letter token (converted to <mi>xy</mi>); $x y$ is two one-letter tokens (converted to <mi>x</mi><mi>y</mi>).

There are a few other places (notably tensor notation, prescripts, etc) where itex departs from standard TeX, in the interest of exploiting certain features of MathML.

Use only right-handed single quotes…

The smart quotes feature turns 'foo' "bar" into ‘foo’ “bar” and … into … (a true ellipsis).

I hope nobody fixes these ‘bugs’, now that I’ve written reams of material that’s adapted to them.

I don’t think they’re bugs. And, for reasons similar to yours, I don’t intend “fixing” them.

Posted by: Jacques Distler on December 1, 2006 7:30 AM | Permalink | PGP Sig

Re: TeXnical Issues

This is a problem.

Posted by: Mike Stay on December 6, 2006 2:35 AM | Permalink

Diagrams

I drew this diagram

(1)P 1(U) p P 1(X) tra U t tra T i T \array{ P_1(U) &amp;\stackrel{p}{\to}&amp; P_1(X) \\ \mathrm{tra}_U \downarrow \;\; &amp;\sim\Downarrow t&amp; \;\; \downarrow \mathrm{tra} \\ T' &amp;\stackrel{i}{\to}&amp; T }

when Toby asked

How did you make this diagram, Urs?

I am using hacks like this one:

\array{
  P_1(U) &\stackrel{p}{\to}& P_1(X)  \\
  \mathrm{tra}_U \downarrow  \;\;  &
  \sim\Downarrow t  &  \;\;  \downarrow
  \mathrm{tra}  \\
  T'  &  \stackrel{i}{\to}  &  T
}
Posted by: urs on December 8, 2006 5:21 PM | Permalink

Re: Diagrams

|array|, eh? Thanks, Urs!

Posted by: Toby Bartels on December 8, 2006 7:52 PM | Permalink

Re: Diagrams

You can achieve a lot of fine-grained control using \array{ \arrayopts{...} ... }. But, in this case, is there anything that couldn’t be achieved with the more familiar \begin{aligned} ... \end{aligned} ?

P 1(U) p P 1(X) tra U t tra T i T \begin{aligned} P_1(U) & \overset{\quad p\quad}{\to} && P_1(X) \\ \mathrm{tra}_U\downarrow & \sim\Downarrow t &&\downarrow \mathrm{tra} \\ T' & \underset{\quad i\quad}{\to} && T \end{aligned}

being produced by

$$
\begin{aligned}
  P_1(U) & \overset{\quad p\quad}{\to} && P_1(X) \\
  \mathrm{tra}_U\downarrow & \sim\Downarrow t &&\downarrow \mathrm{tra} \\
  T' & \underset{\quad i\quad}{\to} && T
\end{aligned}
$$
Posted by: Jacques Distler on December 10, 2006 5:24 AM | Permalink | PGP Sig
Read the post Full Disclosure
Weblog: Musings
Excerpt: A serious MovableType security vulnerability.
Tracked: January 7, 2007 5:19 AM

mimetex public server

I just discovered this: there’s a public server that will take math-mode tex and render it as a gif. (Not that we need it here, but it might be useful to readers.)

Just insert this into your webpage

<img src=”http://www.forkosh.com/mimetex.cgi?c=\sqrt{a^2+b^2}”

    alt=”” border=0 align=middle>

to get this:

Posted by: Mike Stay on February 3, 2007 7:53 PM | Permalink

Re: mimetex public server

Very useful indeed. I wouldn’t want to use up all their bandwidth by using this in by blog, but they might have the cgi available for installation elsewhere.

Posted by: John Armstrong on February 3, 2007 8:39 PM | Permalink

Re: TeXnical Issues

Bruce Bartlett asked me:

I’ve noticed there are a few wizards at the cafe (most notably the hosts and Jacques Distler) who can write down commutative diagrams, insert jpg’s, do accents (like Poincar’e), and even use bold and italic with ease. I don’t know how to do this stuff - it doesn’t seem to work the same way as in LaTeX - so it would be cool if you could explain some of the standard work-arounds.

The key thing to remember is that on this blog, TeX only turns on inside equation environments — that is, inside things like this:

$        $

and this:

$$    $$

and (for numbered equations) this:

 \[     \]

Outside those, your chosen text filter is in charge!

For example:

  • For accents, don’t use TeX commands — they won’t work. Instead, learn how accents and other symbols work in HTML! So, to see

    Poincaré

    don’t write

    Poincar\'e

    Write

    Poincar&eacute;

  • To insert jpgs, you do the usual HTML thing. But, since you’re really using XML, you have to be a good boy and include an ‘alternate text’ for the image in case the one you want is temporarily unavailable. This can be empty. So, to see this:

    (very useful when Urs gets too abstract), you can type this:

    <img src = "http://math.ucr.edu/home/baez/emoticons/cry.gif" 
    alt = ""/>
    

  • I can’t give a complete description of how to draw diagrams, but to get this:
    (1)1 F E B 1 1 F E B 1 \begin{matrix} 1\to& F &\to& E &\to& B & \to 1 \\ &\Vert& &\darr& &\Vert& \\ 1\to& F &\to& E' &\to& B &\to 1 \end{matrix}

    you do this.

    For a longer example, see this post by Urs. Then, see how he did it!

Posted by: John Baez on February 4, 2007 12:07 AM | Permalink

Re: TeXnical Issues

Another route to accents is to learn how your system lets you type in Unicode, rather than remembering all the HTML names.

On my Mac, for example, I can just type option-E followed by a vowel to add an acute accent to that vowel: áéíóú. I’m sure Windows and Unix have something comparable.

Posted by: John Armstrong on February 4, 2007 12:50 AM | Permalink

Commutative diagrams

The key to commutative diagrams (to the extent that we can currently create them here) are the matrix and aligned environments from LaTeX.

\begin{matrix}
...
\end{matrix}

produces an array of centered elements.

\begin{aligned}
...
\end{aligned}

produces an array of elements which are, alternately, right- and left-justified. Typically, it’s used to create multiline equations, but it’s also good for creating a diagram like A h 1 B g 1 g 2 C h 2 D \begin{aligned} A&\overset{\quad h_1\quad}{\to}&&B\\ g_1\darr&&&\darr g_2\\ C&\underset{\quad h_2 \quad}{\to}&&D \end{aligned}

More fine-grained control can be had using the \array{} command (which is not standard LaTeX, but is something itex inherited from WebTeX). Its myriad of options are explained here. Some fancy (but hardly the fanciest) examples of its use can be seen in this post.

Yet fancier diagrams, such as those produced by the dcpic package (which, surely, is the darling of every category theorist) could, in principle, be realized using a mixture of MathML and SVG. While I’d dearly love to implement such a thing, it’s beyond what is currently feasible.

The full list of that is implemented here can be found in the itex command summary.

Posted by: Jacques Distler on February 4, 2007 6:27 AM | Permalink | PGP Sig

Re: TeXnical Issues

OK, this is starting to bug me … and since we’re talking about writing mathematics badly …. What’s with all the primes floating at the feet of their letters, like this?

HH\prime

Shouldn’t the primes be in superscript position, like this?

H H^\prime

Maybe it’s a bug in the interpretation software, but to get the prime in superscript position, you have to do it explicitly, like this (except between dollar signs):

H^\prime

Otherwise the prime isn’t superscripted.

Or is this some weird convention I don’t know about?

Posted by: Tim Silverman on February 23, 2007 10:18 PM | Permalink

Re: TeXnical Issues

Tim, your original example looks just fine to me; your second example is really strained and much too high. You’re not using Internet Explorer, are you …?

Although I’m getting some funny stuff here with Firefox too. Let’s see: * <i>M</i>&prime; should be correct: M′ * <i>M</i><sup>&prime;</sup> should be too high: M * $M'$ should be correct: MM' * $M\prime $ should be too low: MM\prime * $M^{\prime }$ should be correct: M M^{\prime }

Yet I find that the MathML produced by iTeX (in the last three) is wrong as an interpretation of TeX: * <mi>M</mi><mo>'</mo> instead of <mi>M</mi><mo>&#x02032;</mo> * <mi>M</mi><mo>&#x02032;</mo> instead of something that AFAICT cannot be rendered using Unicode characters * <msup><mi>M</mi> <mo>&#x02032;</mo></msup> instead of <mi>M</mi><mo>&#x02032;</mo>

So it appears that your browser is using MathML fonts that compensate for iTeX’s incorrect rendering. Either that, or MathML has redefined the Unicode character U+2032, which iTeX and your fonts have compensated for but my fonts have ignored.

Well, in any case, I think that I will try not to use primes with iTeX here.

Posted by: Toby Bartels on February 23, 2007 10:50 PM | Permalink

Re: TeXnical Issues

Thanks Toby.

So it’s me?

I’m using Firefox.

$M’$ comes out wrong (too low). I guess this is what other people are using.

Is there something I can do about this?

Posted by: Tim Silverman on February 24, 2007 6:50 PM | Permalink

Primes

Bug reports are so nice.

OK. Let’s see what happens now.

$M'$ → MM'

$M''$ → MM''

$M'''$ → MM'''

$M\prime$ → MM\prime

$M^\prime$ → M M^\prime

Better (or, at least, tolerable) now?

Toby says:

$M\prime$ should be too low. … something that AFAICT cannot be rendered using Unicode characters

Well, if it can’t be rendered using Unicode characters, you’re out-of-luck.

Posted by: Jacques Distler on February 24, 2007 9:44 PM | Permalink | PGP Sig

Re: Primes

I’ll tell you what I see, for what it’s worth:

$M’$ low

$M”$ low

$M”’$ high

$M\prime$ low

$M^\prime$ high

My fonts are just a mess.

Posted by: Tim Silverman on February 25, 2007 1:15 PM | Permalink

Re: Primes

Well, if it can’t be rendered using Unicode characters, you’re out-of-luck.

To be fair, probably nobody would want to try it. But somebody might (perhaps thinking of iTeX’s previous treatment of the apostrophe) try $M^\prime $, so that should render correctly (that is, without any actual superscript).

The apostrophe, however, certainly works perfectly for me now! (I’m using Firefox 2, usually on Windows XP.)

Posted by: Toby Bartels on February 27, 2007 2:49 AM | Permalink

SVG in Comments

OK … Why not allow SVG in comments?

A little tweaking was necessary, but you can now do things like this:

Example Young Tableaux in SVG
Some Young Tableaux

or this:

Picture of Trefoil Knot
Framed Trefoil Knot
Posted by: Jacques Distler on April 18, 2007 5:40 AM | Permalink | PGP Sig

Re: SVG in Comments

Excellent! I’ll try one out in the Schur functor thread…

Posted by: John Baez on April 19, 2007 11:45 PM | Permalink

Re: SVG in Comments

Those who would like to use such SV graphics in comments here now and then, but don’t know how to can find more information over at Jacques Distler’s blog.

Jacques suggests to get started to go to the SVG Specification site, copy-and-paste the examples given there and play around with them.

But take care to prefix everything properly.

If in search for SVG inspiration, check out Sam Ruby’s blog which uses it a lot (for decorative purposes, mostly).

Posted by: urs on April 20, 2007 9:19 AM | Permalink

Re: SVG in Comments

There’s always InkScape. But you’ll have to edit the output. In simple cases, it’s easier to just create the figure by hand.

Like this one, for instance:

sphere, cube, tetrahedron
Three Solids
Posted by: Jacques Distler on April 20, 2007 4:18 PM | Permalink | PGP Sig

Re: SVG in Comments

It says “Three Solids”, but I see two!

Posted by: John Baez on April 20, 2007 11:18 PM | Permalink

Legerdemain?

The <svg:desc> element says

sphere, cube, tetrahedron

If you were browsing in a non-SVG-capable browser, you’d see that text, instead of the figure.

Posted by: Jacques Distler on April 21, 2007 2:18 AM | Permalink | PGP Sig

Re: SVG in Comments

I see a beautiful centered picture of a blank space, a cube, and a tetrahedron. I’ve got an up-to-date version of Firefox.

Just a data point…

Posted by: John Baez on April 21, 2007 2:28 AM | Permalink

Re: SVG in Comments

Ulp!

OK. Let’s break it down


Circle with radialGradient fill

versus


Circle with uniform fill

The former was the “sphere” in the previous figure.

A random thought: might you be experiencing bug 372377?

Posted by: Jacques Distler on April 21, 2007 3:29 AM | Permalink | PGP Sig

Re: SVG in Comments

While writing a comment here, I reloaded this page a couple of times. After a while, the SVG output that I saw in the above comments changed.

First everything was displayed beautifully and presumeably the way it was intended.

Then everything appeared in a “negative” kind of way, with bright shades having turned dark, and vice versa.

Reloading even more times then yielded a result as described by John above: the ball with radial gradient fill disappeared entirely, and the cube now appeared with top and front face empty (not filled).

This is for Firefox on Windows (alas).

Posted by: urs on May 2, 2007 4:48 PM | Permalink

Re: SVG in Comments

I believe that was a bug.

Let me know if the problem is fixed for you now.

Posted by: Jacques Distler on May 2, 2007 6:26 PM | Permalink | PGP Sig

Re: SVG in Comments

I’m having similar problems viewing radial gradients in firefox (Linux). I’m trying to track down the bug. If you are experiencing this problem you can help me out by doing the following:

1) Check out my comment in this thread on Jacques’ blog:

Thread

2) Let me know whether or not you see a circle with a radial gradient.

3) If not, click on “Post a New Comment” in that thread. A comment window will pop up. When I do this I see the circle, with radial gradient, properly rendered in this window. Let me know whether or not you see it.

Posted by: Robert McNees on May 9, 2007 10:01 PM | Permalink

Re: SVG in Comments

Yes, exactly the same for me!: Nothing to be seen first, nice graphic visible once I click on “Post a New Comment”.

Firefox 1.5.0.11.

Posted by: urs on May 9, 2007 10:05 PM | Permalink

Re: SVG in Comments

The bug that Jacques described above is definitely part of the problem. The difference between the main thread on his blog and the comments in the pop-up window is that the main thread has a hash # in the address, which kills gradients for some reason.

This explains why you won’t see gradients if you reach this thread via an address containing a ‘#’. However, I don’t think it’s the whole story, because I also experienced a problem with the gradient (in the example above) disappearing after several refreshes. Maybe I accidentally clicked on the ‘permalink’.

In either case, if you are using a recent build of firefox on Windows or Linux (the two cases I have checked), you won’t see gradients in svg for any urls containing a #. Outlines should still appear; it’s just that the example above isn’t stroked.

Posted by: Robert McNees on May 10, 2007 12:57 AM | Permalink

Re: SVG in Comments

You can verify this by appending a # to the end of the url for this page and then clicking reload. Scroll down to this part of the thread and you’ll see that all the gradients have vanished.

Posted by: Robert McNees on May 10, 2007 1:03 AM | Permalink

Re: SVG in Comments

Jacques keeps improving the SVG support here. Now the coding has been simplified a little more. For details, see his SVG Redux.

Posted by: urs on May 2, 2007 4:41 PM | Permalink
Read the post The First Edge of the Cube
Weblog: The n-Category Café
Excerpt: The notion of smooth local i-trivialization of transport n-functors for n=1.
Tracked: May 4, 2007 9:01 PM

Re: TeXnical Issues

I suspect this is probably not something easily fixable in the blogging software but: is there a reason why the comment box appears to be a fixed size in pixels? It’s a bit annoying because I’ve got a largish screen and use larger than default-size fonts but the comment box resolutely sticks to its size, which means I get about 36 characters on a line in the comment box. This makes seeing larger stretches of the text for editing annoying.

Posted by: dave tweed on June 15, 2007 6:10 PM | Permalink

Resizing <textarea>s

I don’t know how many times this question has been asked. The answer is always the same.

All <textarea>s come in a fixed size. This one is 42×10 characters. There is, however, no one size that will suite everyone.

In a browser that supports CSS3 (specifically, the resize property), like Safari, the <textarea> has a thumb that you can drag to resize it.

In other browsers, in particular, in the Mozilla family of browsers, there’s a bookmarklet that allows you to resize any <textarea> with a click of the mouse and a drag.

There are also a couple of Firefox extensions and greasemonkey scripts that do the same thing. (More can be found with a little Googling around.)

Between all of these options, I think you can find something that will work for you. In the longer term, lobby you browser vendor for CSS3 resize support.

[Updated with current, non-broken, links. No, I haven’t tried all permutations and combinations of these solutions. YMMV.]

Posted by: Jacques Distler on June 15, 2007 7:21 PM | Permalink | PGP Sig

Re: Resizing <textarea>s

I’m using Safari and I see no “thumb”.

Posted by: John Armstrong on June 15, 2007 8:12 PM | Permalink

Re: Resizing <textarea>s

I think it only works in Safari 3, which is currently in public beta.

Posted by: Robin on June 15, 2007 8:54 PM | Permalink

Safari

Safari Beta 3 supports the aforementioned <textarea> resizing and inline SVG (as seen in some of the comments above). It’s available for both Mac and Windows.

It doesn’t do MathML, alas.

Posted by: Jacques Distler on June 15, 2007 9:03 PM | Permalink | PGP Sig

Re: Resizing <textarea>s

I’m using Firefox 2.0.0.4 and have installed the extension Resizable Form Fields 0.2.4 and can resize the forms at, for example,the extensions site mentioned above, but I can’t resize the box I’m typing this comment in. Am I missing something?

Posted by: Simon Willerton on June 16, 2007 6:44 PM | Permalink

Resizing textarea

Hooray, the Greasemonkey scripts for resizing the tiny comment box work! I also couldn’t get the Firefox extension(s) to work, but the Greasemonkey one works beautifully. You can work it so it only loads for this site, if you want.

Now I can sit back and comment in peace, with a nice full screen and a HUMUNGOUS textarea :-)

Posted by: Bruce Bartlett on August 28, 2007 2:32 PM | Permalink

Re: Resizing textarea

Okay, for the complete dummies among us:

I have my Firefox here, I grab that Greasemonkey script text – and then I drop it… where?

Posted by: Urs Schreiber on August 28, 2007 2:48 PM | Permalink

Re: Resizing textarea

Steps to install the Greasemonkey script, for Firefox, which allows resizing of the comment box:

1. Install the Greasemonkey add-on for Firefox. You’ll need to restart Firefox again; a little brown monkey should appear in the status bar.

2. Go to the EZ Resize page. Click on “Install this script”. Don’t grab the script text (I did that too :-))… you can, but it’s easier just to click on “Install this script”.

3. You’ll notice that you initially installed it as “runs on *”. This means this script will run on all webpages. If for some reason you just want it to run at at the n-category cafe, right click the brown monkey in the status bar, click “Manage User Scripts”, and change the “Included Pages” part of “EZ Resize” to “http://golem.ph.utexas.edu/*”.

There we go, it should work. Greasemonkey is so cool, you can customize any webpage you want :-)

One further trick - clicking “Ctrl -” makes the fonts smaller in Firefox, so you can see even more text in that comment box.

Posted by: Bruce Bartlett on August 28, 2007 4:00 PM | Permalink

Re: Resizing textarea

Thanks Bruce! Very helpful.

(While Greasemonkey tells me it is unhappy with the version number of my Firefox, I guess I only need to surmount my reluctance to waste my time with these kinds of paper chases of the computer age.)

Now I can sit back and comment in peace, with a nice full screen and a HUMUNGOUS textarea :-)

Great. I am sincerely looking forward to the result of that! :-)

Posted by: Urs Schreiber on August 28, 2007 4:50 PM | Permalink

Re: Resizing textarea

It’s so cool, you can make it infinitely big. So big that even the most ginormous Urs post is humbled to a single line :-)

Now I’m looking forward to getting SVG to work, so you can just “drag-and-drop” images you’ve drawn in, say Inkscape or CorelDraw, onto the comment box, and the inline SVG appears there automatically. The functionality is almost there, I think Jacques just needs to tweak one or two minor things.

Moreover, it will help with all those GIGANTIC diagrams Urs is famous for. You might be able to just drag and drop them onto the comment box, instead of having to include pictures :-)

Posted by: Bruce Bartlett on August 28, 2007 5:24 PM | Permalink

Re: Resizing textarea

So big that even the most ginormous Urs post is humbled to a single line

Right. Like in: the following proof is a one-liner – if you have a large enough scroll bar.

Or as in this fun Hydrogen Atom Scale Model.

inline SVG appears

While I cannot say that I spent a lot of time trying to get around it, I did run into the problem that pasting my SVG code – like that underlying these diagrams – into the nn-Café always made the verifyer complain about exactly the first line of that SVG code, even though I couldn’t see a single difference of that first line with that in Jacques’ example.

That’s why I ended up including these SVGs as .png – which at least harmonizes well with the fact that I am still subject that bug which prevents the SVG from being displayed in the first place.

I am sure this can all be dealt with somehow. But I haven’t really tried hard, so far.

Posted by: Urs Schreiber on August 28, 2007 6:00 PM | Permalink

Re: Resizing textarea

Mmm… interesting. How do you make your diagrams Urs? Everyone seems to have various workarounds, and I’m interested in finding nice solutions. When I’m typing up things in LaTex, I use a convoluted scheme involving CorelDraw, various font packages, and then I export everything as .eps files, which get \includegraphics’ed in the Latex file.

Advantages : draw in CorelDraw; text on images done in CorelDraw directly using Latex fonts (no pstricks workarounds). Disadvantages: Text doesn’t always come out right; can’t include \xymatrix{}, and so on.

Posted by: Bruce Bartlett on August 28, 2007 6:42 PM | Permalink

Re: Resizing textarea

How do you make your diagrams Urs?

The black&white xy-pic sort of things I produce by first creating an ordinary LaTeX file and then running that through the simple but convenient TeX To GIF script.

Those SVG graphics, though, were done in InkScape and then exported as png (or as plain source code, but that didn’t work here, as I said).

Posted by: Urs Schreiber on August 28, 2007 6:55 PM | Permalink

Re: Resizing textarea

That’s great, Bruce!

Everyone should do what Bruce said. It’s very helpful.

Posted by: John Baez on August 29, 2007 2:05 PM | Permalink

Re: Resizing textarea

Perhaps we should adjust this Greasemonkey script, so that it has clear variables “PermanentXSize=” and “PermanentYSize=”; setting these variables once-for-all then makes the comment textarea remain that size permanently. At the moment it returns to the old size after you’ve previewed something.

The next thing we need is for Mike Stay or Jacques to write up a script for colourful blog conversations, in the spirit of Colorful Tabs, the best Firefox add-on since… Firefox!

It can be called “ColorfulCafé”, the first n-category skin! I was thinking of a system where each conversation thread in a particular post is assigned a pale background color, and as new replies are added to a particular thread, the backgrounds for the older conversations get paler and paler and the recent additions to a thread get brighter and brighter.

Posted by: Bruce Bartlett on August 29, 2007 3:15 PM | Permalink

Re: Resizing textarea

Wonderful, Bruce. Thanks. Now my text can breathe freely.

Posted by: Tom Leinster on August 29, 2007 4:49 PM | Permalink

Re: Resizing <textarea>s

Hooray! In Google Chrome, one can adjust the textarea for the comments automatically, as in Safari.

Posted by: Bruce Bartlett on September 5, 2008 8:29 PM | Permalink

Re: Resizing <textarea>s

But the maths doesn’t display properly! Does anybody know how to fix this?

Posted by: Jamie Vicary on September 5, 2008 10:52 PM | Permalink

XML parsing problem

I’m having a problem viewing one of the n-Cafe pages, and unless my memory’s playing tricks on me this problem didn’t exist a couple of days ago.

When I go to this address, my browser (Firefox) gives me the following error message:

XML Parsing Error: no element found
Location: http://golem.ph.utexas.edu/category/2007/02/this_weeks_finds_in_mathematic_5.html
Line Number 3961, Column 26:
in which some unfinish
----------------------^

I’ve certainly viewed this page before on this browser on this computer, and I’m almost certain I did so a couple of days ago. It works fine with Konqueror. Any thoughts?

Thanks.

Posted by: Tom Leinster on July 15, 2007 11:43 PM | Permalink

Re: XML parsing problem

It’s fine for me with Firefox 2. Is there a way to work out what happens at line 3961?

Posted by: David Corfield on July 16, 2007 8:06 AM | Permalink

Re: XML parsing problem

Yeah, I’m using Firefox 2 now and it works. I think it’s an older (though not particularly old) version of Firefox on the computer that I was using when I had problems.

Also, my certainty that I’d viewed that page before with that computer has diminished slightly since I wrote the comment above…

You can easily find where the problem takes place by searching for the text “in which some unfinished”. As to why there’s a problem, goodness knows.

Posted by: Tom Leinster on July 16, 2007 7:31 PM | Permalink

Re: XML parsing problem

Weird stuff occasionally happens on this setup. It’s mostly beyond my understanding.

Chris Hillman keeps pushing for us to switch to a wiki that allows LaTeX, for example something like what Garrett Lisi is now using. One advantage is that readers would not need to download special fonts. It could also be more robust.

I’m lazy about making changes like this.

Posted by: John Baez on July 17, 2007 10:03 AM | Permalink

Re: TeXnical Issues

Whenever I post comments, it takes a long time to preview them, and a really long time (about 15 seconds) to post them… does everybody find this? It can’t just be that I’m the other side of the Atlantic to the server. The only time this is a real problem is when I’m repeatedly previewing/editing to get something to come out right, but other blogs don’t seem to have this problem, so it seems silly to put up with it. Also, multiply 15 seconds by the number of comments posted here in the last year, and you’ll get a good number of hours!

Posted by: Jamie Vicary on September 11, 2007 9:53 PM | Permalink

Re: TeXnical Issues

I timed that last submission: 15 seconds is a spectacular underestimate. It took 2 minutes 18 seconds!!

Posted by: Jamie Vicary on September 11, 2007 9:59 PM | Permalink

Re: TeXnical Issues

… and that one took 1 minute 19 seconds, so there’s clearly quite a lot of variability.

Posted by: Jamie Vicary on September 11, 2007 10:02 PM | Permalink

Re: TeXnical Issues

Yes, my experience is about the same as yours (using a computer in the UK): I’d guess that posting takes a minute or two, and previewing 15 or 30 seconds. It doesn’t really bother me though.

Posted by: Tom Leinster on September 12, 2007 12:24 AM | Permalink

Re: TeXnical Issues

Jamie Vicary wrote:

When I post comments, it takes a long time to preview them, and a really long time (about 15 seconds) to post them… does everybody find this? It can’t just be that I’m the other side of the Atlantic to the server.

Even in the US, it always takes longer than I’d like — about 15 seconds, maybe. In Paris and Vienna it was about the same. Jacques Distler has discussed the problem in another thread.

But right now, here in the Sheffield University mathematics department computer room, it’s taking a lot longer: over a minute. So, I suspect the university is using some sort of cheap, low-quality internet access. It would be in keeping with the cheap, low-quality workstation I’m on. Maybe you’re suffering from something like that?

Posted by: John Baez on September 12, 2007 10:57 AM | Permalink

Re: TeXnical Issues

So, I suspect the university is using some sort of cheap, low-quality internet access … Maybe you’re suffering from something like that?

Well, I’m posting from home rather than university, but I’ve got a blazing broadband connection, and I can view pages from the Cafe very quickly. It seems totally bizarre to me that it would take 2 minutes to post from the UK, but only 15 seconds in the States, using an otherwise speedy connection… I can’t think of a reason for that at all.

Perhaps submissions are deliberately being delayed by an evil organisation, determined to prevent inter-continental category theory from flourishing!

Posted by: Jamie Vicary on September 12, 2007 12:01 PM | Permalink

Re: TeXnical Issues

Could the ‘slows down posting’ warning mentioned here have any connection to this issue?

Posted by: Jamie Vicary on September 12, 2007 1:11 PM | Permalink

Re: TeXnical Issues

But right now, here in the Sheffield University mathematics department computer room, it’s taking a lot longer: over a minute. So, I suspect the university is using some sort of cheap, low-quality internet access.

Heh - quite possibly! But I doubt that’s what’s causing the slowdown. I have the same experience - it often takes over a minute for some of my posts to go through - even though I post them from my own home, and not from the Sheffield computers.

I have some vague notion that the Big Slowdown started at a certain well-defined point in time, which I think was between six months and a year ago.

Before that things were humming along. Nowadays it’s a case of submitting a post, going downstairs to make dinner, and coming back after dinner to see if the post has gone through. Jacques has explained the likely cause - but can there be a mysterious “UK” effect, and did it start at some definite point in the past? These are the questions of our time.

Posted by: Bruce Bartlett on September 12, 2007 5:01 PM | Permalink

Slowdown

Before that things were humming along. Nowadays it’s a case of submitting a post, going downstairs to make dinner, and coming back after dinner to see if the post has gone through. Jacques has explained the likely cause - but can there be a mysterious “UK” effect, and did it start at some definite point in the past? These are the questions of our time.

I can’t think of why the UK would be singled out.

Another possible source of slowdown is that, before your comment is posted, the server does a DNSBL lookup of your IP address.

It could be that one of the DNSBL servers is wonky and fails to respond in a timely fashion. That would cause a delay (for everyone, not just UK customers). I don’t think that’s the problem; I periodically check the servers that we use, and they all seem to respond rapidly whenever I check them manually.

Still, as an experiment, I’ve disabled this particular anti-spam measure to see if it makes a difference for y’all.

Posted by: Jacques Distler on September 12, 2007 8:14 PM | Permalink | PGP Sig

Re: Slowdown

Test…

Posted by: Jamie Vicary on September 12, 2007 10:22 PM | Permalink

Re: Slowdown

That took 2 minutes and 18 seconds; oddly consistent with my first test. If John Baez’s estimate is representative, then it certainly does seem as if the UK is being singled out.

Posted by: Jamie Vicary on September 12, 2007 10:43 PM | Permalink

Re: Slowdown

Another test.

Posted by: David Corfield on September 13, 2007 7:58 AM | Permalink

Re: Slowdown

2 minutes 2 seconds from Yorkshire (UK). Hmmm.

Posted by: David Corfield on September 13, 2007 8:02 AM | Permalink

Re: Slowdown

The delays on my system (continental Europe, in case that matters ;-) are not on the order of minutes, even though it does take a couple of seconds. Longer than on other blogs I know.

Not sure if it is relevant, but those engaged in benchmarking the nn-Café submission delays might try to compare on their system the following observation, which might help pin down the source of the delay:

I find that long before the edit window gets to the point that it admits that the comment was posted, the comment in fact is already visible on the blog.

I think

- first the comment appears on the page of the given blog entry

- then, after some further time, its presence is accounted for in the counter beneath the entry’s excerpt on the main page “Followups (nn)”.

- Then, at the very end of the delay period, the comment also appears in the list “recent comment” on the title page.

Maybe those experiencing the very long delays could check which of these three intervals consumes most of the delay.

Posted by: Urs Schreiber on September 13, 2007 5:41 PM | Permalink

Re: TeXnical Issues

Back here in Greenwich the delays seem roughly comparable to those in Sheffield. So, I guess Sheffield isn’t to blame. I don’t remember the delays being so bad before.

I’ll return to California on Tuesday September 17th, and then I’ll see how the delays in the US compare to those in the UK.

Jamie wrote:

Perhaps submissions are deliberately being delayed by an evil organisation, determined to prevent inter-continental category theory from flourishing!

More likely it’s Dr. Evil, taking revenge against our attacks against PRISM.



Posted by: John Baez on September 13, 2007 10:54 PM | Permalink

Re: TeXnical Issues

I’ve found that my posts actually get published within a second, but the server takes a very long time to come back and tell me so. After hitting submit on the popup compose-your-post page, refresh the thread page and you’ll see it’s already there long before the popup page gets around to telling you it’s finished.

Posted by: Mike Stay on September 14, 2007 4:57 PM | Permalink

Re: TeXnical Issues

Surely it’s a bug in the script, then? Perhaps we should report it to whoever wrote the code.

Posted by: Jamie Vicary on September 16, 2007 4:58 PM | Permalink

Re: TeXnical Issues

Your observation accords with the email notification theory as to the cause of the delay.

Just FYI, posting the previous comment took 30 seconds. A bit slow (I’d be happier with ~15 seconds), but perfectly acceptable.

Posted by: Jacques Distler on September 16, 2007 7:19 PM | Permalink | PGP Sig

Re: slowness

I also find that previewing and posting comments takes a long time (about 40 seconds to preview this comment), but that (when posting) the comments appear long before the web page comes back (supporting the email notification theory, although that shouldn’t affect previews). However, for me the speed seems to depend at least partly on how many comments there currently are on the particular entry I’m commenting on. An entry like this one, with lots of comments, takes a really long time, whereas a newer one with hardly any comments is pretty zippy. Any theories about that?

Posted by: Mike Shulman on October 9, 2009 9:37 PM | Permalink | PGP Sig

Re: TeXnical Issues

Here’s another technical question. At the bottom of the comments pages, there is a list of ‘access keys’, which it seems should be used together with the ALT key. But, they don’t seem to work. How can I turn them on? I’m using the latest version of Firefox on a Windows XP machine. They look like they could be quite useful.

Posted by: Jamie Vicary on September 16, 2007 5:16 PM | Permalink

Accesskeys in Firefox

I believe you need Alt-shift-key. See here.

Posted by: Jacques Distler on September 16, 2007 7:11 PM | Permalink | PGP Sig

Re: TeXnical Issues

Urs drew a diagram which drew a reply by Jacques that led to a discussion of the rendering of diagrams, math symbols and the like. I’m moving these comments to a more suitable place — namely, here. I don’t want to squelch the discussion, just put it someplace where people who need to know these things can find it!

Robin Houston wrote:

Just fyi, it isn’t better on my computer, where (in your rendition) the lower ‘arrow’ is rendered as a question mark, presumably because I don’t have a font that contains the character you used.

(The upper arrow appears to be Unicode CLOCKWISE TOP SEMICIRCLE ARROW (U+21B7), which – although it’s probably as close as one can get with a single Unicode character – doesn’t really look appropriate in any of the fonts I have installed.)

Mike Stay wrote:

Just fyi, it isn’t better on my computer

Mine either. It’s a box with “293B” in it.

Surely with all the different markup environments you’ve got now, you could add one more: a pstricks environment for drawing all the nice pictures and diagrams that everyone’s so fond of here.

If it’s easier to adapt someone else’s code, then this could be a nice starting point: IllustRender plugin for WordPress.

Dan Christensen wrote:

My two cents: how about providing a link that produces a pdf file showing how a comment is supposed to look? Despite lots of effort on my part, much of the mathml doesn’t render well for me. Most of the time I can guess what was intended, but it would be nice to be able to occasionally see a pdf so I could be sure.

Jacques Distler wrote:

In the future, all these font problems will be a thing of the past.

The lower arrow is \curvearrowbotright=U+293B=⤻. The upper (ugly) arrow is &#x21B7;=U+21B7=↷. On my system, both are provided by the CODE2000 font.

If it’s easier to adapt someone else’s code, then this could be a nice starting point: IllustRender plugin for WordPress.

I’m not sure what problem that solves.

Inkscape and inline SVG ought to be as nice as Illustrator and a plugin to render Illustrator files to PNG. But Inkscape is open-source.

Despite lots of effort on my part, much of the mathml doesn’t render well for me.

Jason Blevins put together some very nice test pages for itex2MML on his Instiki wiki. Maybe you could use those to narrow down what doesn’t (and what does) work for you.

Posted by: John Baez on October 8, 2007 8:35 PM | Permalink

Re: TeXnical Issues

IllustRender plugin for WordPress.

I’m not sure what problem that solves. Inkscape and inline SVG ought to be as nice as Illustrator and a plugin to render Illustrator files to PNG.

Did you look at it? It has nothing to do with Adobe Illustrator; it uses metapost and postscript. As for what problem it solves, it allows you to have TeX in your diagrams, unlike inline SVG.

But a pstricks environment (please please!) would let you use nodes for commutative diagrams while still allowing arrows to curve nicely.

Posted by: Mike Stay on October 8, 2007 9:36 PM | Permalink

Fonts, glyphs, and pixies

I have finally gotten the fonts to work on my machine and so would like to record what I did for posterity. This may not be the slickest route to a working mathml but it works for me.

Thanks to a couple of hints from Jacques (quoted by John above) and the information on the test pages, I was able to work out that the missing glyphs were in the CODE200X (X = 0 or 1) fonts.

For the record, my system is a Unix-based one which is not Mac OS X and my browser uses the mozilla engine (specifically, FreeBSD running Galeon) with xft-fonts enabled. I have downloaded the following fonts:

  1. TeX fonts
  2. Mathematica 4.1 fonts
  3. CODE2000
  4. CODE2001

I also downloaded CODE2002 but that’s probably overkill. I found a copy of the Symbol font (Type 1) from an installation of acroreader (may be worth installing acroreader to get this font even if you never use it!). I put them all into my ~/.fonts directory and hey presto it all worked.

Because I got the Symbol font from acroreader, I didn’t need to do anything funny from these instructions.

I also downloaded the MT Extra TTF as suggested in the above instructions (note: you don’t have to supply an email address) but I couldn’t find a program on my system to extract the fonts from the file (cabextract would probably work but I didn’t think it worth installing just for these fonts).

So, my ~/.fonts directory contains:

 CODE2000.TTF
 CODE2001.TTF
 CODE2002.TTF
 Symbol
 cmex10.ttf
 cmmi10.ttf
 cmr10.ttf
 cmsy10.ttf
 math1___.ttf
 math1b__.ttf
 math1m__.ttf
 math1mb_.ttf
 math2___.ttf
 math2b__.ttf
 math2m__.ttf
 math2mb_.ttf
 math3___.ttf
 math3b__.ttf
 math3m__.ttf
 math3mb_.ttf
 math4___.ttf
 math4b__.ttf
 math4m__.ttf
 math4mb_.ttf
 math5___.ttf
 math5b__.ttf
 math5m__.ttf
 math5mb_.ttf

And finally I can understand what you are all writing … I mean, finally I can read what you are all writing. Comprehension might take a little longer. Sadly, that’s not yet available as a download.

Of course, once the STIX fonts are released, this misery will all ”softly and silently vanish away”.

Posted by: Andrew Stacey on October 10, 2007 12:10 PM | Permalink

Re: TeXnical Issues

Ah, installing the Code2000 font helped a huge amount. I hadn’t seen this mentioned on any of the instructions I had looked at. Andrew Stacey’s message should help a lot of users. I didn’t install Code2001 or Code2002, because from their descriptions they don’t seem likely to be useful.

It would still be quite useful to have a “view as pdf” link on each entry to check how things are working. Sometimes I honestly can’t tell if what I’m seeing is correct!

I tried the itex test suite and most things work. Things that still don’t display correctly are:

- dashrightarrow, curvearrowbotright, circlearrowleft and circlearrowright don’t work.

- The mathbb, mathfrak and mathcal fonts only have a few entries.

- The “aligned” equations have too much space before the alignment point.

- The \: and \; commands don’t produce any more space than \, does.

- The \over{x}{y} test just displays \over{x}{y} literally.

Posted by: Dan Christensen on October 12, 2007 2:56 AM | Permalink

Fonts

Ah, installing the Code2000 font helped a huge amount.

For me, it supplies a half-dozen miscellaneous symbols (mostly arrows) which would otherwise be missing. Not a big deal. (Certainly, not as important as the other mathematical fonts.)

I hadn’t seen this mentioned on any of the instructions I had looked at.

Please look at the browser page on my Instiki site, for the best information I have.

Andrew Stacey’s message should help a lot of users. I didn’t install Code2001 or Code2002, because from their descriptions they don’t seem likely to be useful.

Code2001 is essential. It contains several key Math alphabets: blackboard bold (𝔸), fraktur (𝔄) and calligraphic (𝒜). These alphabets are all in Unicode Plane 1, and there aren’t any other font options, currently.

There are a few letters from these alphabets in the BMP, which you will, doubtless, find in other fonts. But the full alphabets are only available in Code2001.

Code2002, by contrast, covers Unicode Plane 2 and is, for present purposes, completely useless.

  • dashrightarrow, curvearrowbotright, circlearrowleft and circlearrowright don’t work.

All supplied by Code2000.

  • The mathbb, mathfrak and mathcal fonts only have a few entries.

As I said, you need Code2001.

    • The “aligned” equations have too much space before the alignment point.

I believe that’s related to this bug, specifically, that the columnspacing attribute is not respected..

  • The \: and \; commands don’t produce any more space than \, does.

They do for me, although the difference is subtle.

  • The \over{x}{y} test just displays \over{x}{y} literally.

That test was broken. It’s fixed now.

Posted by: Jacques Distler on October 12, 2007 3:52 AM | Permalink | PGP Sig

Re: Fonts

Dan wrote:

I hadn’t seen this mentioned on any of the instructions I had looked at.

Jacques replied:

Please look at the browser page on my Instiki site, for the best information I have.

At the risk of seeming like a complete luser, could you update the FAQ at the top of this page to provide a link to your Instiki site? At the moment, the only obvious link points to the mozilla page which only tells you about the basic mathml fonts. This is the most obvious place to put this information and as Urs gets more and more ambitious with his diagrams then getting the fonts right will help us all keep up.

I may be wrong in this, but I feel that I’m fairly savvy about computers and if it took me this long and this much hassle to find out about the CODE200X fonts then I can imagine many people giving up and going off to get Real Jobs™ rather than staying and getting hooked on mathematics.

Posted by: Andrew Stacey on October 12, 2007 8:44 AM | Permalink

Re: Fonts

I’ll second that. I also feel reasonably savvy about computer things, and I just learned about code200x fonts. To be sure, I’m also pretty lazy about computer things, but then so are most people.

Posted by: James on October 12, 2007 11:43 AM | Permalink

Re: Fonts

Please look at the browser page on my Instiki site, for the best information I have.

Thanks, that page definitely deserves to be linked to from the FAQ at the top of this page!

I see that you don’t suggest installing cmsy10 and cmex10; I used to get warning messages saying that they were missing, but they went away, probably because I added the line you suggested to user.js. So I guess there’s no advantage to having those installed?

I didn’t install Code2001 or Code2002, because from their descriptions they don’t seem likely to be useful.

Code2001 is essential. It contains several key Math alphabets: blackboard bold (𝔸), fraktur (𝔄) and calligraphic (𝒜).

Thanks for pointing that out. The descriptions I read didn’t mention this.

dashrightarrow, curvearrowbotright, circlearrowleft and circlearrowright don’t work.

All supplied by Code2000.

Those symbols didn’t show up for me yesterday, even though Code2000 was working (for example, dashleftarrow was fine). Today, after installing Code2001, all those symbols are there. It may be unrelated to Code2001, but I didn’t change anything else.

The \: and \; commands don’t produce any more space than \, does.

They do for me, although the difference is subtle.

You’re right.

The \over{x}{y} test just displays \over{x}{y} literally.

That test was broken. It’s fixed now.

Yep, thanks. I’m now happy to say that all of the tests from that page seem to work for me!

Dan

Posted by: Dan Christensen on October 13, 2007 1:30 AM | Permalink

Re: Fonts

This little discussion may explain why Chris Hillman has never been able to read math here on his UNIX machine — and why I’ve never been able to do it on mine, either.

I hope he (and I) try it again.

I’ll add the information to the FAQ on top of this page. And, I hope Jacques adds a suitable link to the “MathML” blurb on the top right of our main page.

Posted by: John Baez on October 13, 2007 1:40 AM | Permalink

Re: Fonts

I see that you don’t suggest installing cmsy10 and cmex10; I used to get warning messages saying that they were missing, but they went away, probably because I added the line you suggested to user.js.

Among other things, that line controls which fonts Mozilla will nag you about when you visit a MathML-enabled page.

The default value (which you have overridded) asks for the Mathematica fonts, the Computer Modern fonts and for the Symbol font.

So I guess there’s no advantage to having those installed?

On a Mac, they’re, apparently, harmful. In any case, it’s redundant to have both the Mathematica and the Computer Modern fonts installed.

(for example, dashleftarrow was fine…)

That glyph is available from lots of different fonts. A useful tool you might consider is Unicode Checker. If you search for “dashed left arrow”, you’ll find that U+21E0 is present in lots of fonts, including Lucida Grande, which you certainly have installed on your system.

Today, after installing Code2001, all those symbols are there. It may be unrelated to Code2001, but I didn’t change anything else.

Probably some font-cache getting rebuilt. Or, maybe it was leprechauns…

Yep, thanks. I’m now happy to say that all of the tests from that page seem to work for me!

Those are supposed to be comprehensive tests of itex. If there’s something Jason missed, that’s broken for you here, let us know, and we’ll add a test for it.

To my knowledge, the only feed reader that supports MathML is Liferea.

Thanks for pointing it out. I’ve now read about it but haven’t tried it yet. Installing using fink has about 30 dependencies, so I thought I’d ask one more question before installing it:

To be honest, I’ve only heard, 2nd hand, that it now does. I’ve asked a friend, involved in the project, to confirm.

Great, glad to hear there is a mechanism for this. Do you know if liferea supports this?

I’ve seen it as a feature request for several feed readers. Dunno who’s actually gone ahead and implemented it. If you have a favourite feed reader, request that they implement RFC 4685, and tell ‘em why (so you can view blog comment feeds as threaded).

This little discussion may explain why Chris Hillman has never been able to read math here on his UNIX machine

My impression was that Chris’s beef with MathML ran deeper than the need to install a few fonts.

— and why I’ve never been able to do it on mine, either.

I hope he (and I) try it again.

If you do, I encourage you to add RedHat/Fedora instructions to that page.

Posted by: Jacques Distler on October 13, 2007 3:53 PM | Permalink | PGP Sig

Re: Fonts

I justed switched from using OSX to Ubuntu on my Macbook Pro, and thought I’d mention that most of the itex tests pass after I followed the simple instructions here.

Dan

Posted by: Dan Christensen on November 6, 2007 2:47 AM | Permalink

RSS reader suggestions

I find it awkward to follow the n-category cafe from its web-based interface, as it doesn’t keep track of what I have and haven’t read. (I’m much better at following mailing lists and newsgroups, but then you don’t have the itex formatting…)

So I often use Google Reader to keep up. But it doesn’t display the MathML as well. For example, spacing is wrong (e.g. equations are flush left), the fonts used are different (but still ok), and superscripts often appear as regular-sized symbols at the usual height. My first question is whether this can be fixed. Maybe Google is not providing the stylesheet?

My second problem is that the comments aren’t threaded in Google Reader. As far as I can tell, Google Reader doesn’t offer any kind of threading. I tried a different web-based RSS reader which did offer threading, but it based it on the subject lines, and since they all start with the name of the author, this made the threading not work. Is there a way around this?

Or is there a better RSS reader I should use?

How do others keep up with the volume of posts in different topics?

Posted by: Dan Christensen on October 12, 2007 3:04 AM | Permalink

PDF and RSS reader suggestions

It would still be quite useful to have a “view as pdf” link on each entry to check how things are working.

You’ve asked for this several times.

The answer is no.

But I guess you won’t be satisfied until I explain why.

The only way I know of to obtain a PDF file is to convert your comments to LaTeX and run them through PDFLaTeX.

Doing that conversion is complicated. We mostly have that working in Instiki. There, we had it relatively easy, because

  1. There was only one markup dialect to deal with (Markdown+itex), instead of the choice of multiple text filters that you have here.
  2. Much of the heavy lifting was done by Maruku, which has a LaTeX output option. Creating one for even just the Markdown filter in MovableType would be ridiculously hard.

Even so, we had to write some really nontrivial LaTeX macros, and still, they are not complete. (I would be most appreciative if you were willing to volunteer to write the LaTeX macro to implement itex’s \array command.)

Moreover, I don’t consider the laTeX output robust enough for automated PDF generation. It’s perfectly fine if there’s a human around to correct the odd glitch. But for automated PDF generation, you want something absolutely bullet-proof, which this isn’t.

That said, Jason re-enabled automatic PDF generation on his Instiki wiki (I had disabled it). I wish him luck, but I won’t do that in my distribution because

  1. As I said, the LaTeX generation isn’t sufficiently bulletproof.
  2. PDFLaTeX is slow.
  3. It introduces yet another external dependency (a complete teTeX or TexLive distribution) which which will not be present on many webhosts.

On the subject of feed readers

So I often use Google Reader to keep up. But it doesn’t display the MathML as well.

To my knowledge, the only feed reader that supports MathML is Liferea.

My second problem is that the comments aren’t threaded in Google Reader. As far as I can tell, Google Reader doesn’t offer any kind of threading. I tried a different web-based RSS reader which did offer threading, but it based it on the subject lines, and since they all start with the name of the author, this made the threading not work. Is there a way around this?

Find a feed reader that supports RFC 4685, the Atom Threading Extension, rather than trying to hack together threads, based on the subject line. That’s why there are %$#@ Internet Standards in the first place.

Posted by: Jacques Distler on October 12, 2007 5:28 AM | Permalink | PGP Sig

Re: PDF and RSS reader suggestions

It would still be quite useful to have a “view as pdf” link on each entry to check how things are working.

You’ve asked for this several times.

I had asked for it once before. I repeated the question since I thought it might have been missed.

The answer is no.

But I guess you won’t be satisfied until I explain why.

The only way I know of to obtain a PDF file is to convert your comments to LaTeX and run them through PDFLaTeX.

Yes, that sounds complicated, especially with the different markup languages. I had actually imagined something that works from the html+mathml that is produced, using a standard renderer (e.g. mozilla’s engine) in a controlled environment with all the appropriate fonts installed. Not that I think this would be easy!

So I often use Google Reader to keep up. But it doesn’t display the MathML as well.

To my knowledge, the only feed reader that supports MathML is Liferea.

Thanks for pointing it out. I’ve now read about it but haven’t tried it yet. Installing using fink has about 30 dependencies, so I thought I’d ask one more question before installing it:

My second problem is that the comments aren’t threaded in Google Reader.

Find a feed reader that supports RFC 4685, the Atom Threading Extension

Great, glad to hear there is a mechanism for this. Do you know if liferea supports this?

Thanks for all of your help!

Posted by: Dan Christensen on October 13, 2007 1:58 AM | Permalink

Liferea

Sam Ruby confirms that Liferea:

  1. supports MathML and SVG
  2. supports RFC 4685, the Atom Threading Extension

Enjoy.

Posted by: Jacques Distler on October 19, 2007 7:48 PM | Permalink | PGP Sig

Re: Liferea

I just saw your message and installed liferea. MathML does indeed seem to work, but I can’t see anything in the menus or documentation about threading. Maybe I’m just misinterpreting you, but I was hoping to see the comments in tree order, based on which comments they are replying to.

Dan

Posted by: Dan Christensen on November 6, 2007 2:42 AM | Permalink

Re: Liferea

Since I never got threading working in liferea, I’ve just started trying something new: I have a python program that reads the provided feeds and splits them into separate feeds based on the subject line and other info. The result is that each new non-comment blog entry starts a new feed, and all of the comments related to that post show up in that feed. My program also outputs the list of feeds in “OPML” format, which liferea understands, so liferea automatically subscribes to these new feeds.

I’ve only been using this system for a couple of days, but to me it is a big improvement.

Two questions for Jacques:

1) Would adding one feed per main post be something you might consider doing? Someone could subscribe to the main entries feed, and when he or she reads something that they want to follow the discussion on, they could click on a provided link to subscribe to the associated feed.

2) I’ve noticed that the comments.atom file only contains 1 to 2 days worth of comments. If I continue my home-brew system, do I need to ensure that I poll this file once a day, or is there somewhere else I can find data that goes back farther?

Thanks,

Dan

Posted by: Dan Christensen on June 16, 2008 3:25 AM | Permalink

Per-entry comment feeds

2) I’ve noticed that the comments.atom file only contains 1 to 2 days worth of comments. If I continue my home-brew system, do I need to ensure that I poll this file once a day, or is there somewhere else I can find data that goes back farther?

The comment feed contains the last 20 comments, regardless of when they were made. This number is a tradeoff between polling frequency and file length, to optimize bandwidth.

Typical clients poll with a frequency somewhere between once an hour and once a day. The file should be long enough to contain all the comments since the last time the client polled, but not so long that the client is retrieving a large number of old comments, just to get the one new one that they haven’t seen before.

A properly-written client caches what they retrieve, and supports ETags and/or the Last-Modified header.

How many comments you cache is entirely up to you. You can cache the last hundred or the last thousand or whatever number you want. The only reason the number of comments contained in the feed should be an issue for you is that you need to adjust your polling frequency so that the last 20 contains all new comments since the last poll.

If I were to set out to do what you are doing (take the n-Category Café comment feed, and split it into one feed per entry), I would implement that as a Venus filter.

1) Would adding one feed per main post be something you might consider doing?

I’ve struggled over that. The basic issue is: when should a “per-entry” comment feed go away?

In my experience, there are people (or ‘bots or whatever) who will subscribe to every feed URL you advertise. And they will never unsubscribe; they will continue requesting that resource, at regular intervals, forever.

So, unless I want to service a monotonically increasing number of requests for a monotonically increasing number of feeds, forever, I need to expire per-entry comment feeds after some period of time:

  • a week after the last comment on that entry was made?
  • a month?
  • what about an entry which hasn’t received comments for a while, but which suddenly receives a new one? Should I resurrect the per-entry comment feed?
  • what if that comment was spam?

These questions, plus some MovableType-related implementation issues, that are not of interest to anyone else but me, have kept me from implementing such a thing. But no one has expressed much of an interest before…

Posted by: Jacques Distler on June 16, 2008 2:59 PM | Permalink | PGP Sig

Re: Per-entry comment feeds

Jacques wrote:

> Dan wrote:
>
>> 2) I’ve noticed that the comments.atom file only contains 1 to 2
>> days worth of comments.
>
> The comment feed contains the last 20 comments, regardless of when
> they were made. This number is a tradeoff between polling frequency
> and file length, to optimize bandwidth.

Oh, I’m surprised by that. I can easily imagine 20 comments getting
posted in an hour, so even with hourly polling an RSS reader may
miss entries. Not to mention what happens when I bring my laptop
on a plane or something like that. Do you think you could guarantee
that you keep items for at least 24 hours?

> If I were to set out to do what you are doing (take the n-Category
> Café comment feed, and split it into one feed per entry), I would
> implement that as a Venus filter.

Planet Venus looks interesting. It uses the same parser I did
(Universal Feed Parser), but so far I don’t see how it can be used to
make a variable number of output feeds. I’ve asked on the mailing
list.

>> 1) Would adding one feed per main post be something you might
>> consider doing?

> I’ve struggled over that. The basic issue is: when should a
> “per-entry” comment feed go away?

> […]

> So, unless I want to service a monotonically increasing number of
> requests for a monotonically increasing number of feeds, forever, I
> need to expire per-entry comment feeds after some period of time:

You raise some interesting points, but I think any reasonable expiry
policy would be fine.

> But no one has expressed much of an interest before…

Well, I don’t understand how people read the n-category cafe. If you
read it via the web, nothing keeps track of the comments you’ve seen.
You can poll each thread you are interested in, but that is time
consuming, and some threads get interesting posts for quite a while.
And if you read it in an rss reader, when you have a busy week you come
back to several hundred comments all jumbled together. (Sorting by
subject doesn’t help, since the title of the thread isn’t at the start
of the subject.)

I guess I’m just an nntp/mailing list sort of guy, where the
information gets sent to me.

Posted by: Dan Christensen on June 17, 2008 3:05 AM | Permalink

Re: Per-entry comment feeds

Oh, I’m surprised by that. I can easily imagine 20 comments getting posted in an hour, so even with hourly polling an RSS reader may miss entries.

On some blogs, perhaps, but not here.

I looked at the past two months of data; the n-Category Café averages 15 comments per day (albeit, with large scatter). Polling once/hour is almost certainly safe. Polling once/day will, on occasion, miss stuff.

Do you think you could guarantee that you keep items for at least 24 hours?

I’ll look into it.

It would involve hacking MT’s internals. The <mtentries> container tag accepts both a lastn and days attribute. But the <mtcomments> container tag accepts only lastn.

Well, I don’t understand how people read the n-category cafe. If you read it via the web, nothing keeps track of the comments you’ve seen. You can poll each thread you are interested in, but that is time consuming, and some threads get interesting posts for quite a while. And if you read it in an rss reader, when you have a busy week you come back to several hundred comments all jumbled together. (Sorting by subject doesn’t help, since the title of the thread isn’t at the start of the subject.)

Information overload is a problem. I don’t think there’s a “silver bullet” solution.

For the particular purpose of catching up on the comments to a particular entry, the view chronologically button is invaluable.

But, you’re right, one of the advantages of feedreaders (and, for that matter NNTP clients) is that they keep track of what you’ve read.

Posted by: Jacques Distler on June 17, 2008 8:23 AM | Permalink | PGP Sig

Re: PDF and RSS reader suggestions

It would still be quite useful to have a “view as pdf” link on each entry to check how things are working.

You’ve asked for this several times.

The answer is no.

How about a “view comment in separate window/tab” box? That way I can use my browser’s print-to-p{s,pdf} facility to export a given comment, but don’t have to slurp the whole discussion if I don’t want to.

Andrew

Posted by: Andrew Stacey on January 7, 2008 11:22 AM | Permalink

Printing a comment

Why is that preferable to Firefox’s “Print Selection” option?

The latter lets you print an arbitrary contiguous segment of the page, not just those segments I think you might be interested in printing.

Posted by: Jacques Distler on January 7, 2008 2:38 PM | Permalink | PGP Sig

Re: Printing a comment

Basically because I was completely unaware of Firefox’s “print selection” capability! Now that I know of it, I partially withdraw my suggestion.

Only partially because it is just possible that there are people who look at this blog who don’t use firefox. Me, for example. Fortunately the “print selection” is also supported in Galeon and, on a quick test, did what I expected it to.

Also only partially because it may be that some others don’t know that some web browers can, using “print selection to file”, export a segment of a webpage to a P{S,DF}. Now they do.

Thanks for the help!

Andrew

Posted by: Andrew Stacey on January 11, 2008 8:09 AM | Permalink

Re: Printing a comment

Also only partially because it may be that some others don’t know that some web browers can, using “print selection to file”, export a segment of a webpage to a P{S,DF}. Now they do.

Almost all modern browsers have a “Print Selection” function. It’s unfortunate that many people don’t know about it.

But I take this as a teaching opportunity. People will be much happier, in the end, knowing about this capability of their browser which

  1. works on any website (not just this one) and
  2. works better than anything I could kludge together on the server-side.

Sometimes it’s better to recast a problem (“I’d like to be able to print an individual comment on this website.”) in more general form (“I’d like to be able to print an arbitrary contiguous segment of the page on any website.”) before attempting to find a solution.

This is true in Mathematics as well.

Thanks for the help!

De rien!

Posted by: Jacques Distler on January 11, 2008 3:39 PM | Permalink | PGP Sig

RSS reader in IE7

Hi,

I recently “upgraded” to IE7 and, to my surprise, actually seem to like it better than Firefox so far. I imported all my feeds from Firefox to IE7 and of the hundreds of feeds I have, The n-Category Cafe is the only one whose article title links do not seem to work.

Specifically, I am subscribed to the comments feed.

I am not emotional about this because I can read most of the content from the RSS feed, but if I want to click through to see the equations formatted correctly, I cannot. Just FYI.

Cheers

PS: Sorry if this has already been discussed, but the “Search” feature also does not seem to work for me. It complained about the XML.

Posted by: Eric on February 25, 2008 12:33 AM | Permalink

Putting author name at top of posts?

Would it be possible to put the author’s name at both the top and bottom of each message they write? As it is now, as I scroll down the page I can’t tell who’s voice I should imagine while reading a comment until I scroll down far enough to see the author.

Posted by: Dan Christensen on October 12, 2007 3:09 AM | Permalink

Re: Putting author name at top of posts?

I’ll second this, too.

Posted by: James on October 12, 2007 11:47 AM | Permalink

Re: Putting author name at top of posts?

Actually, I quite like being forced to guess. For instance, it’s really easy to spot Urs and John. David’s quite easy too, I find. Guessing right gives me a satisfied glow.

Posted by: Tom Leinster on October 13, 2007 7:15 PM | Permalink

Re: Putting author name at top of posts?

He said, “Son, I’ve made a life out of readin’ people’s faces,
Knowin’ what their cards were by the way they held their eyes,
And if you don’t mind me sayin’, I can see you’re out of aces,
And for a taste of your whiskey I’ll give you some advice.”

Posted by: Bruce Bartlett on October 13, 2007 10:49 PM | Permalink

Re: Putting author name at top of posts?

I’m a bit late to this conversation but I’ll third Dan’s suggestion of wanting the author’s name at the top.

We could club together and buy Tom a light box to make up for his lost satisfied glow?

Posted by: Eugenia Cheng on November 25, 2007 7:47 PM | Permalink

Re: Putting author name at top of posts?

I’ll chuck in a few kroner for that light box. Can we please have author names at the top of entries? It really would help me figure out who’s saying what rather than having to keep scanning down to the bottom and back up again.

Posted by: Andrew Stacey on January 4, 2008 3:07 PM | Permalink

Clocks changed

There’s a slightly disorienting bug this week. The time stamp on posts are an hour out for me. I assume that the server asks my machine what time zone I’m in and then calculates them appropriately. Unfortunately it doesn’t seem to be aware that the clocks went back at the weekend.

PS Is it possible to put a link to this thread somewhere accessible? I can never remember how the title of the post is spelt and so can never find it.

Posted by: Simon Willerton on October 31, 2007 10:51 PM | Permalink

Re: Clocks changed

Actually, in the USA the clocks didn’t change this weekend, though many computer systems didn’t get the message. Thanks Congress.

Posted by: John Armstrong on October 31, 2007 11:45 PM | Permalink

Finding This Thread

Well, if we keep finding problems and posting comments about them, this thread will always be on the main-page sidebar.

Posted by: Blake Stacey on November 26, 2007 1:56 AM | Permalink

Finding This Thread

Okay, I’ll post a comment here to keep this thread on the main-page sidebar — at least long enough for Simon to remember the word ‘TeXnical’.

(It’s a pun on ‘TeX’. Get it? It’s supposed to be memorable.)

Seriously, I could pester Jacques Distler to deal with this — he’s the only one able to adjust such things — but he’s currently involved in labors of a more herculean sort. And, he seems to be making progress! All the horrid bugs that infest the back end of this blog — which you, dear readers, hardly ever notice — seem to be gone!

So, I should let him rest a while before urging him to make it easier to find this thread.

Posted by: John Baez on November 27, 2007 7:30 PM | Permalink

Re: Finding This Thread

The “Random Past Entries” section of the sidebar is now only pseudo-random. A bit of a kludge, but hopefully it will satisfy.

All the horrid bugs that infest the back end of this blog — which you, dear readers, hardly ever notice — seem to be gone!

I’m glad to hear things seem to be improving. I can’t say that I’ve definitively “solved” anything.

Indeed, for several hours this afternoon, the server had gotten itself into a severely muddled state in which DNS was no longer working, notification emails were no longer being sent, and the only solution was, ultimately, to reboot the machine (at around 16:26 CST).

Seems to be another Leopard bug. Sigh

Posted by: Jacques Distler on November 28, 2007 3:39 AM | Permalink | PGP Sig

Re: TeXnical Issues

I’ve just recently upgraded my Safari, and it doesn’t seem to have remembered all the old cookies properly. In particular, it doesn’t seem to “remember” me when I post comments.

So I deleted the cookies from this site and the problem hasn’t resolved itself. It seems that the site isn’t sending the cookie to my browser. Has anyone else noticed anything about the site’s cookies behaving oddly?

Posted by: John Armstrong on November 22, 2007 1:04 AM | Permalink

Re: TeXnical Issues

Nope! Lots of other problems, but not that!

Posted by: John Baez on November 22, 2007 1:21 AM | Permalink

Re: TeXnical Issues

Regarding cookies and Safari. Cookies from this site are not working for me when I use Safari. I’ve asked it to remember me but it isn’t doing so.

My settings under Safari are to accept cookies only from the site that I navigate to. Are the cookies being sent from somewhere other than ‘golem.ph.utexas.edu’?

I won’t change my cookie setting for obvious reasons but, given John’s comment above, it may be something to do with Safari itself.

Next time I’m in front of a Linux machine I’ll see if a mozilla-based browser has the same problem.

Posted by: Andrew Stacey on January 4, 2008 3:03 PM | Permalink

Cookies in Safari

Cookies are broken in Safari in XHTML. Safari (the current version) does support SVG, however.

So, rather than back off, and send text/html to Safari, we’re hoping the Apple people get off their asses and fix their Javascript DOM support under XHTML.

Personally, I would consider a nice Cocoa-based browser, like Camino. Or Firefox/Mac. Both support MathML, which Safari doesn’t (and likely won’t, for the foreseeable future).

Posted by: Jacques Distler on January 4, 2008 4:41 PM | Permalink | PGP Sig

Re: TeXnical Issues

Why do all the equations look like gobbledygook?

Maybe you do not know maths, so all equations look like gobbledygook.

Hear hear!!

*crawls back to hole*

Posted by: Beans on December 17, 2007 12:13 PM | Permalink

Composition is broken

The composition operator \circ renders as a degree symbol, presumably to accommodate ^\circ. However, on this blog at least, composition turns up far more often than temperature. A workaround would be to fix the mathml rendering of &#9702; [◦], which currently displays [&#9702;] instead.

Posted by: Mike Stay on January 10, 2008 10:38 PM | Permalink

Re: Composition is broken

Howso broken?

The MathML named entity, &compfn;, maps to U+2218 (∘) the Unicode character, “RING OPERATOR.”

That’s what the itex command, \circ maps to, as well.

At least on my machine, the font used in equations looks better than the one selected for running text:

ABA \circ B

But that’s one of those crazy browser-dependent things …

Posted by: Jacques Distler on January 11, 2008 5:07 AM | Permalink | PGP Sig

Re: Composition is broken

Every font on my computer renders x2218 the same, as a degree symbol. I found a page that said it shouldn’t, but have no idea how to fix it. sigh

Posted by: Mike Stay on January 11, 2008 5:13 PM | Permalink

Re: Composition is broken

So, how do they render U+00B0 (°), which is the degree sign in Unicode?

[You could, of course, try out the Stix fonts, but I’m tickled that U+2218 (\circ) renders as a degree sign on your machine.]

Posted by: Jacques Distler on January 11, 2008 7:33 PM | Permalink | PGP Sig

Re: Composition is broken

It’s a slightly larger degree sign.

I hacked my way around the problem by telling my greasemonkey script to replace any occurrence of 2218 in a formula with an image of the proper symbol.

Posted by: Mike Stay on January 11, 2008 9:59 PM | Permalink

Problem with recent comments

I have two problems with “recent comments” : the time is wrong (the minutes are “01” for each comment), and many comments don’t appear in the list : for example, for “Classifying spaces for 2-groups”, the last comment which appear in the list is from John Baez and begins with «David wrote: Pathetically simple …», but there were 13 new comments since this one! I already noticed before missing comments in the list, but not so much. I use Firefox on Mac OS X.

Thank you!

Posted by: Mathieu Dupont on January 28, 2008 9:01 AM | Permalink

Re: Problem with recent comments

Some bug with the comments is bugging us currently. Until this has been fixed, and generally, notice:

  • When composing longer comments, be sure to save a copy in some text editor.
  • When you receive an error message after submitting your comments, chances are that your comment actually did go through, and only one of the numerous auxiliary tasks the software tries to carry out failed, like updating the “recent comments” list. You can check this by reloading the corresponding entry, choosing “view chronologically” and see if your comment is there.

When you notice that an annoying lot of recent comments are not being listed in the “recent comments” list, you can drop us blog hosts a notice, and we can force the blog to rebuild itself.

Of course, ideally none of this should be necessary, but while the bug persists, proceeding this way will make life easier.

(Saving a separate copy of a comment is generally a good idea, though. I always have one editor window open for saving things that I am about to send away.)

Posted by: Urs Schreiber on January 28, 2008 2:58 PM | Permalink

RSS feed problem?

Is the RSS feed for recent comments playing up at the moment?

In the last two days I’ve started getting only non-useful things in the feed like

‘The post “Blah blah” comments on “Blah blah blah”’

where “Blah blah” could be “Klein-2-Geometry X” and “Blah blah blah” could be “Klein-2-Geometry VIIIS”.

I don’t seem to be seeing any actual comments.

Posted by: Simon Willerton on April 17, 2008 11:40 AM | Permalink

Re: RSS feed problem?

Thanks, Simon! Should be fixed now.

Posted by: Urs Schreiber on April 17, 2008 5:35 PM | Permalink

Re: RSS feed problem?

It is, thanks.

Posted by: Simon Willerton on April 17, 2008 5:52 PM | Permalink

Re: TeXnical Issues

I’m having some trouble getting text to come out right (i.e. in sensible font, size, etc) inside SVG, due most likely to my almost non-existent XML skills.

What do I need to do to get normal-looking text? Please be painfully explicit …

Posted by: Tim Silverman on May 31, 2008 4:07 PM | Permalink

Re: TeXnical Issues

I’m having some trouble getting text to come out right (i.e. in sensible font, size, etc) inside SVG, …

First of all, Firefox’s SVG implementation is, to put it mildly, a work in progress. Before you blame deficiencies in your XML skills, take my advice and try a nightly build of Firefox, and see if things display correctly there.

Anyway, here are some simple examples to get you started.

<svg xmlns="http://www.w3.org/2000/svg" width="10em" height="10em" viewBox="0 0 160 160">
  <circle cx="80" cy="80" r="70" fill="#AAF" stroke="#000"/>
  <text x="20" y="80">svg:text example</text>
</svg>

produces:

svg:text example

<svg xmlns="http://www.w3.org/2000/svg" width="5em" height="5em" viewBox="0 0 80 80">
  <g fill="#AFA" stroke="#000" stroke-width="1">
    <rect width="30" height="30" x="10" y="10"/>
    <rect width="30" height="30" x="40" y="10"/>
    <rect width="30" height="30" x="10" y="40"/>
  </g>
  <foreignObject width="20" height="20" x="15" y="10">$\alpha_1$</foreignObject>
  <foreignObject width="20" height="20" x="45" y="10">$\alpha_2$</foreignObject>
  <foreignObject width="20" height="20" x="15" y="40">$\alpha_3$</foreignObject>
</svg>

produces:

α 1\alpha_1 α 2\alpha_2 α 3\alpha_3

More examples (including embedding SVG in equations) can be found here and here and here.

Posted by: Jacques Distler on May 31, 2008 7:35 PM | Permalink | PGP Sig

Re: TeXnical Issues

Thanks for your reply, Jacques.

I’d be happy to download an overnight build of Firefox, but the place you point to for the MacOSX version no longer exists! And I can’t find anything that seems similar by poking about the website. You wouldn’t happen to know where the new location is?

The code snippets you gave don’t display at all in my browser as they are, since it doesn’t seem to recognise “em” in the size. Replacing the em specifications with just numbers gives the following results:

The first snippet (with text) displays fine if I save it into a file and load the file into Firefox; but if I put it in the Café comment box, the Preview shows the blue circle but not the text.

The second snippet obviously won’t render directly in Firefox; if I Preview it through the Café, I get the boxes but no symbols, although the page source contains mathml that looks OK. If I copy the source into a file, and load that into a window in Firefox, I again get the boxes but no symbols, although if I just copy and paste the embedded mathml, that renders fine on its own.

In this post, the text rendered but looked weird. That made me wonder if there was some default text style for the Café that I could refer to/tweak/override, but I had no idea what it was, or how to find it, or use it (on account of aforementioned XML ignorance).

I guess you probably just want to tell me to download the latest Firefox build, rather than specifying bugs in detail—well, I would if I could! But I also live in hope that this has a workaround or is my fault in some way….

Posted by: Tim Silverman on May 31, 2008 10:54 PM | Permalink

Re: TeXnical Issues

I’d be happy to download an overnight build of Firefox, but the place you point to for the MacOSX version no longer exists!

The place exists; the filename has changed. Try here (and download the .dmg file).

since it doesn’t seem to recognise “em” in the size.

That’s right. Sizing in “em”s is a new feature.

The second snippet obviously won’t render directly in Firefox;

Again, <foreignObject> (which allows you to embed MathML in SVG) is a new feature.

I don’t know why you should have seen any difference between a standalone file and a Café comment. Obviously, the markup I suggested “works” in the comments here. Either your browser supports it (in which case, it should render in either context) or it doesn’t.

In any case, try the Firefox Nightly Build that I linked to, and then we can talk further.

Posted by: Jacques Distler on May 31, 2008 11:28 PM | Permalink | PGP Sig

Re: TeXnical Issues

Only the filename’s different? Sheesh, I’m an idiot, and I apologise.

Anyway, I’m now running the new version, and it’s much better. Thanks!

Posted by: Tim Silverman on June 1, 2008 1:09 AM | Permalink

Re: TeXnical Issues

In this post, the text rendered but looked weird. That made me wonder if there was some default text style for the Café that I could refer to/tweak/override, but I had no idea what it was, or how to find it, or use it (on account of aforementioned XML ignorance).

SVG <text>, alas, gets rendered in whatever default font the browser decides upon. You could tweak this by setting the font-family="..." attribute. But you may find that math looks better as … math. Which means using SVG <foreignObject> (and itex equations).

This may be overkill for the simple fractions in your Stern-Brocot Trees, but for anything more complicated …

Posted by: Jacques Distler on June 1, 2008 4:34 AM | Permalink | PGP Sig

Re: TeXnical Issues

I actually wanted to handle those fractions using mathml, but I didn’t know how! So you conveniently answered my next question before I’d even asked it.

Thanks again.

Posted by: Tim Silverman on June 1, 2008 12:52 PM | Permalink

Re: TeXnical Issues

Well, now you see why I am so eager to have people migrate to the Mozilla Nightly’s and the STIX fonts.

Among other things, we do want everyone to be able to see the equations in your diagrams.

Posted by: Jacques Distler on June 1, 2008 3:40 PM | Permalink | PGP Sig

Re: TeXnical Issues

Yes. I want to start using diagrams a lot more. It’s frustrating having handwritten notes full of pages of diagrams but, when I post, only posting descriptions of them, which are incomprehensible unless the readers draw them out themselves (or have a stupendous visual imagination). So I will be very disappointed if I post a lot of diagrams but most people can’t see them properly.

Posted by: Tim Silverman on June 1, 2008 5:08 PM | Permalink

Re: TeXnical Issues

I feel your pain.

It will be so nice when the next release of Firefox is out, and it becomes reasonable to assume that most people (at least, most people with an interest in the content hereabouts) are using a browser with good support for MathML+SVG.

Fortunately, one of the virtues of open source is that that day won’t be terribly far in the future. Another virtue is that, in the interim, there are development versions available which do implement the features we want, and we can encourage as many people as possible to use them.

I’ve been using SVG on my blog since 2003. I kinda regret some of the hacky compromises that I made, to work around the limitations of browsers at the time. But there’s no way I’m going back to revise all those old posts. In composing similar content, today, I’m strongly inclined to do so in a fashion that I won’t feel the urge to go back and revise my posts when browser technology improves.

Posted by: Jacques Distler on June 1, 2008 5:31 PM | Permalink | PGP Sig

Re: TeXnical Issues

Tim wrote:

I will be very disappointed if I post a lot of diagrams but most people can’t see them properly.

Your nice comment on Stern–Brocot trees looked okay, I think… but Jacques’ examples come out looking completely blank to me!

So: what the heck is a ‘Firefox Nightly Build’? Do the elves leave a different version of this browser under the Christmas tree at night?

By the way — I hope you noticed my request: I’d like to put up a version of your Stern–Brocot comment as a ‘guest post’, so more people can see it. Is that okay?

Posted by: John Baez on June 2, 2008 7:51 AM | Permalink

Re: TeXnical Issues

So: what the heck is a ‘Firefox Nightly Build’? Do the elves leave a different version of this browser under the Christmas tree at night?

Pretty much.

It’s compiled (by the elves) from the latest committed version of the Firefox source repository once every 24 hours.

The upside: it has all the latest bugfixes and improvements.
The downside: it has all the latest bugs and regressions.

As far as MathML and SVG support go, the benefits of the former far outway the hazards of the latter.

I do all my browsing using the nightly builds. (If you’re running a recent Firefox, you’ll notice that it is set to periodically update itself with the latest version. In the case of a nightly build, this means updating to the latest nightly.) While not bug-free, I almost always find it a less buggy and more satisfying experience than running the current release version.

Posted by: Jacques Distler on June 2, 2008 9:04 AM | Permalink | PGP Sig

Re: TeXnical Issues

John wrote:

Do the elves leave a different version of this browser under the Christmas tree at night?

Ah, now I understand the fairy-tale: he was an open source shoemaker.

And furthermore:

I hope you noticed my request

No. But if you emailed it to me, the reason may be that when I moved to the nightly build of Firefox, it lost a few settings or cookies (I guess) including my email address in the Café comment form—and when I typed it in again, there was a typo. Should be fixed now.

And most pleasingly,

I’d like to put up a version of your Stern-Brocot comment as a ‘guest post’, so more people can see it. Is that okay?

I would be delighted, and honoured.

Posted by: Tim Silverman on June 2, 2008 10:55 PM | Permalink

Re: TeXnical Issues

I will be very disappointed if I post a lot of diagrams but most people can’t see them properly.

I have no choice and am stuck with IE. Consequently, I cannot read 80% of the equations anyway.

Posted by: Eric on June 2, 2008 3:23 PM | Permalink

Re: TeXnical Issues

How could you be ‘stuck with Internet Explorer’? It took me ten minutes to download and install Firefox on my Windows machine, and I’m a complete idiot when it comes to computers. Ever since, I’ve rarely used Internet Explorer.

(Certain weird websites are only compatible with IE, but not many.)

Posted by: John Baez on June 2, 2008 5:13 PM | Permalink

Re: TeXnical Issues

To install ANY software, you need “admin” privileges. I do not have admin privileges at work (where I spend most of my time). Firefox is not allowed unfortunately.

Posted by: Eric on June 2, 2008 5:57 PM | Permalink

Re: TeXnical Issues

Ouch.

Posted by: John Baez on June 2, 2008 6:23 PM | Permalink

Re: TeXnical Issues

Could I make the following request for the cafe? The in-built search function is reeaal slow, and pretty shoddy in general. (I believe this observation is originally due to James Dolan pic!) Perhaps a prominent link to a Google-style search through the golem.ph.utexas.edu/category site would make everyone happier? For instance,

“3-morphism site:golem.ph.utexas.edu/category”

immediately brings up the good stuff. Also, perhaps some smallish section on how to read this blog for beginners, containing hints on RSS software and so on which make the blog-experience more palatable. This gimungous TeXnical Issues section is not really digestible for a first-time visitor - nor is it an easy link to spot if someone goes to the homepage for the first time.

In fact, I’m curious to know how various n-cafe readers actually read this blog; perhaps I can learn something. I use the Sage reader and click on “recent comments”, which brings up the opening paragraphs of all the recent comments. I seldom actually go to the main homepage… sometimes that means I am blind to new posts.

Posted by: Bruce Bartlett on July 11, 2008 10:19 PM | Permalink

Re: TeXnical Issues

I used to use Sage, but now I use Google reader.

Posted by: Eric on July 11, 2008 11:22 PM | Permalink

Re: TeXnical Issues

Bruce wrote:

In fact, I’m curious to know how various nn-Café readers actually read this blog; perhaps I can learn something.

You won’t learn much from me. Whenever I’m feeling bored, I use Firefox to go to the main page and look for new posts and then new comments. My moderatorial superpowers also ensure that I get an email whenever anyone comments on any thread I initiated.

I agree that the search engine is slow. But what I really dislike (and I think this is what bugged Jim) is that by default it searches only blog entries, so I almost have to click on ‘search comments’ to get the results I want.

Posted by: John Baez on July 12, 2008 5:45 PM | Permalink

Re: TeXnical Issues

Something strange has been going on for hours now — whenever I try to post a comment through my usual internet connection, I get the message “Your comment submission failed for the following reasons: You are not allowed to post comments.”

I have now connected to my university’s VPN, and it works! This seems pretty strange. Hopefully the patrons aren’t banning me deliberately :/

Posted by: Jamie Vicary on August 26, 2008 9:44 PM | Permalink

Re: TeXnical Issues

Yeah, we don’t let people brandishing daggers into this café.

Seriously, I’m not sure what the problem is. I’m sorry it’s occuring! If it persists, please ask Jacques if your IP address has somehow been put onto a ‘banned’ list. This happens sometimes, even for people who aren’t evil spammers.

Posted by: John Baez on August 27, 2008 3:37 AM | Permalink

Re: TeXnical Issues

I really like the visitor locations map, the big version of which is here… I hadn’t noticed it before! It’s interesting to see where the activity is. What’s the big blob just to the south-west of the Great Lakes? There are big voids in Russia and the Middle East, which surprises me, and apparently no readers in Canada, which is surely impossible. Also, not many hits from the west-coast USA! John, what are your students doing all day??

Posted by: Jamie Vicary on September 11, 2008 2:51 PM | Permalink

Re: TeXnical Issues

That’s odd. When I click on that link you gave, Jamie, I just see a single dot somewhere in the path of Ike, which is our host Jacques Distler. From there however I can go to ‘map with small clusters’ and see more hits. No Canada, but several US west coast.

Posted by: David Corfield on September 11, 2008 3:47 PM | Permalink

Re: TeXnical Issues

I see Quebec City, Montreal and Toronto, and it looks like Seattle and Vancouver get combined on the big cluster view.

Posted by: Aaron Bergman on September 11, 2008 4:03 PM | Permalink

Re: TeXnical Issues

No Canada

Actually, I think there are a bunch of Canadian hits, but they’re hard to see because most Canadians live so close to the US border. If you switch to the smaller cluster view, you can see lots of dots that look like they are in south-western Ontario (where I am).

Is there a way to zoom in and/or get a higher resolution map?

Posted by: Dan Christensen on September 13, 2008 8:11 PM | Permalink

Re: TeXnical Issues

It sounds like we’re seeing different things, so here are my views as of 15:00 GMT on the 11th of September 2008:

and with smaller clusters:

Click on these maps for larger views.

Jamie writes:

What’s the big blob just to the south-west of the Great Lakes?

Chicago, with the University of Chicago (home of Peter May) and Northwestern University (home of Ezra Getzler)? And/or maybe Madison Wisconsin, with the University of Wisconsin?

I wish these maps had more resolution. In fact I wish you could combine the information with Google Earth, and zoom in to see the home of everyone reading this blog!

(I’m just kidding, since that would be quite invasive, but I wonder if it’s technically feasible.)

Also, not many hits from the west-coast USA! John, what are your students doing all day??

Hey, somebody has got to write my papers while I blog all day!

Seriously, I think my grad students have always tended to be too busy to spend much time reading the nn-Café. In a way that’s sad, since it means they don’t get as much of a sense of the grand scheme of things. But on the other hand, maybe it’s good that they’re focused on projects instead of wallowing in erudite chatter.

Right now Chris Rogers and I are trying to understand ‘quantization’ and ‘projectivization’ as adjoint functors. Alex Hoffnung and Christopher Walker and I are writing an easy introduction to groupoidification. And John Huerta… well, right now he’s visiting friends in Portland, but he’s mainly writing a paper called ‘The Algebra of Grand Unified Theories’.

Would it be better if they spent more time at the nn-Café? I don’t know. I really wonder about that.

Posted by: John Baez on September 11, 2008 6:10 PM | Permalink

Re: TeXnical Issues

Would it be better if they spent more time at the nn-Café?

Yes! :-)

Also, I find it a pity that Jeffry Morton is not (cross-)posting his stuff here. it is so closely related to lots of the things we are talking about. While I track his blog and comment there, I am sure there would have been much more discussion of his posts had they appeared here.

Clearly there is an upper bound to the topics and contributors one can sensibly have on a single blog, but for the time being I think the synergy effect of concentrating related dicussion here outweighs any disadvantages this may have.

Posted by: Urs Schreiber on September 11, 2008 9:42 PM | Permalink

Re: TeXnical Issues

I’ve invited both Todd Trimble and Jeff Morton to post more stuff here. I think their posts could add a lot of life to the n-Café, and they’d probably get more comments here.

I’m also trying to lure some other people here, but I don’t want to name them prematurely and scare them off! I think a lot of people are a bit hesitant to join the fun. I don’t completely understand why, but I think some people are afraid they’d have to submerge their identity under ours, instead of being themselves.

Posted by: John Baez on September 13, 2008 1:55 AM | Permalink

Re: TeXnical Issues

Todd’s also written some great expositional posts which would be very suited for the Wiki. How’s that going by the way?

Posted by: David Corfield on September 13, 2008 9:53 AM | Permalink

Re: TeXnical Issues

David wrote:

How’s that going by the way?

It’s all set up, apparently — I just can’t get it to work. Whenever I try to make a change in the page I get an error message saying

403 Forbidden

You must have Javascript on to submit this form.

I’ve done a bunch of obvious things. In particular, I’m using Firefox 3.0.1. and I have Javascript enabled; the same problem happens to me with Explorer. But the guy who set up the page doesn’t experience this problem!

Hey, it looks like Todd Trimble managed to edit this page successfully!

Let’s see if you can do it, David. Go this page, click on ‘Edit Page’, and see if you can edit it. Just follow your nose; it’s supposed to be really easy.

I think the name of this wiki is incredibly cool: Lisa invented it. Much better than ‘The Wizard’s Workshop’. I’ll be pissed off if I’m the only one in the universe who can’t get it to work.

Any idea what could cause this error message in Instiki, Jacques?

Posted by: John Baez on September 13, 2008 7:18 PM | Permalink

Re: TeXnical Issues

Yes, I can edit it, but, like Simon, no Latex.

Posted by: David Corfield on September 13, 2008 10:08 PM | Permalink

Re: TeXnical Issues

John wrote in part:

I’ll be pissed off if I’m the only one in the universe who can’t get it to work.

Well, cheer up, because I can’t get it to work either!

I’m also using Firefox 3.0.1, more specifically ‘Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9.0.1) Gecko/2008070208 Firefox/3.0.1’ (found under Help -> About in the Firefox menus). And I have JavaScript on (as proved by the success in posting this comment). But I have cookies blocked, if you think that that might make a difference.

Posted by: Toby Bartels on September 14, 2008 3:56 AM | Permalink

The wiki

I just wrote in part:

But I have cookies blocked, if you think that that might make a difference.

Yep, that was it. (See my edit to the test page.) So you (whoever ‘you’ is) should change the HTTP 403 message to say that Javascript and cookies (from math.ucr.edu) are required (or at least say that such cookies are required when the browser already has Javascript).

Or of course, you could fix it to no longer require Javascript or cookies, but maybe that would take more work or interfere with some anti-spam plans.

Posted by: Toby Bartels on September 14, 2008 4:16 AM | Permalink

Re: TeXnical Issues

Had a quick play at the nLabTM^TM but couldn’t get the itex maths to display as maths in HTML. (The ‘TeX’ source looks okay though.)

Posted by: Simon Willerton on September 13, 2008 9:30 PM | Permalink

Re: TeXnical Issues

Though I live in Perth, on the west coast of Australia, my ISP is based in Melbourne, on the east coast. So I suspect that nothing I do is going to create even the smallest blip on your map in my actual location.

Posted by: Greg Egan on September 13, 2008 11:26 AM | Permalink

Re: TeXnical Issues

Jacques: any chance we could get the ability to edit comments?

We don’t want people to be able to edit each others’ comments, so one idea is OpenID. But we really don’t need anything so complicated.

When a comment is posted, you could give the poster a random number, rather like the tag you get when you check your coat, that he could present when it’s time to edit the comment.

Posted by: Mike Stay on November 10, 2008 6:02 PM | Permalink

Re: TeXnical Issues

There are various reasons I’m not eager for people to start editing their comments. For example: if there’s some sort of debate going on, it’s no fair to go backwards in time and change ones argument to make ones discussion partner look foolish.

However, in my role as benevolent oligarch, I routinely go back in time and change my own comments to remove annoying typos and errors before anyone has commented on them. I also do this for other people.

(After someone has comment on a mistake, it’s not really fair to remove the mistake — unless they agree to have their comment deleted!)

I was going to do this for some of your recent mistakes, and maybe I’ll do it now.

Posted by: John Baez on November 10, 2008 8:52 PM | Permalink

Re: TeXnical Issues

Regarding changing comments, there are some blog sites (this one as a random example) use some sort of cookie magic to allow the author to revise a comment precisely when no-one has yet responded to it. (If someone posts a reply as whilst you edit your edit upload is refused.)

However, I find that allowable editing tends to be used only be very rarely so it’s probably not worth the effort.

Posted by: bane on November 11, 2008 9:20 AM | Permalink

Re: TeXnical Issues

I’m not sure how else to report this, so I’ll throw it in here. After recently updating Safari to 3.2, both Safari and OmniWeb are displaying this message while rendering the front page of this site:

“This page contains the following errors:
error on line 1672 at column 880: Entity ‘cacute’ not defined
error on line 1674 at column 953: Entity ‘cacute’ not defined
error on line 1675 at column 1055: Entity ‘cacute’ not defined
Below is a rendering of the page up to the first error.”

Posted by: Richard on November 17, 2008 4:20 AM | Permalink

Re: TeXnical Issues

The &cacute; problem seems to have reappeared. At the time of writing, Safari inserts the following in a pink box at the top of the café’s front page:

This page contains the following errors:
error on line 564 at column 5457: Entity 'cacute' not defined
Below is a rendering of the page up to the first error.

It’s true too: there’s a &cacute; in the Recent Comments box.

Posted by: Robin Houston on February 24, 2009 9:21 PM | Permalink

Re: TeXnical Issues

It looks as if there’s something massaging the entity name into a hexadecimal code when it shows up almost everywhere on the site – but not in the Recent Comments.

For example, this post is full of c-acutes, but they’re all specified with hexadecimal codes (at least by the time it reaches my browser). But somewhere deep in the basement of the café the title must be represented in a form that still includes the named entity &cacute;, and it’s not being translated when it shows up in the Recent Comments box.

Posted by: Greg Egan on February 25, 2009 1:17 PM | Permalink

Re: TeXnical Issues

Richard’s computer wrote:

error on line 1672 at column 880: Entity ‘cacute’ not defined

I know that &eacute; is standard HTML that produces this: é. (I see this as an e with an acute accent.)

Let’s see what &cacute; produces. It produces this: ć. (I see this as a c with an acute accent. What do you see? Have I just poisoned this page for you?)

Posted by: John Baez on November 17, 2008 4:56 PM | Permalink

Re: TeXnical Issues

You didn’t actually produce an entity &cacute;; Jacque’s scripts turned this into &#x107; when you posted your comment. So I’m sure that you didn’t break anything, but that proves nothing.

By my understanding, there is no such thing as &cacute;, although Jacque’s intelligent scripts figured out what you probably meant. So if &cacute; does appear on the main page, then that would be an error, one that would crash some browsers.

(It appears to occur in the RDF code just before “Baković on Bigroupoid 2-Torsors”. I think that it’s a bug for a browser to crash on that, since that is technically just a comment in the overall XML structure. It does not appear in that post itself!)

Posted by: Toby Bartels on November 17, 2008 9:17 PM | Permalink

Re: TeXnical Issues

I was getting the same errors as Richard when I upgraded to Safari 3.2, but they went away shortly after he commented on them. When I looked at the home page’s source code, I noticed that there was (now?) no use of the element &cacute;, but a numerical code of &#x107; instead which displays correctly in Safari as c with an acute accent. The same is true of John Baez’s comment above. So maybe Jacques worked some magic to make this happen?

(Of course it’s very silly that Safari responded so dramatically to an unknown entity name.)

Posted by: Greg Egan on November 17, 2008 9:15 PM | Permalink

&cacute;

So maybe Jacques worked some magic to make this happen?

Any sufficiently advanced technology is indistinguishable from magic.

By my understanding, there is no such thing as &cacute;

There is. In the MathML DTD, just not in the HTML 4 DTD.

Safari 3.2, apparently, is persnickity about such things, in ways that previous version of Safari were not(?).

Posted by: Jacques Distler on November 17, 2008 10:33 PM | Permalink | PGP Sig

Re: &cacute;

Jacque wrote:

Toby wrote:

By my understanding, there is no such thing as ć,

There is. In the MathML DTD, just not in the HTML 4 DTD.

Aha! I only know what's valid in HTML and XML, but I forgot that this was XHTML+MathML!

Presumably, any “entity id” here is valid. That includes &cacute;.

Safari 3.2, apparently, is persnickity about such things,

I believe that a browser is required to be persnickity about such things if it is to qualify as XML compliant (at least when served Strict, as in Robin's comment).

Posted by: Toby Bartels on November 18, 2008 10:16 PM | Permalink

Re: TeXnical Issues

I tried sticking “&cacute;” in an HTML file and looking at it with Safari 3.2, which displayed it as “&cacute;” rather than “ć”, but gave no error message. So I’m not sure what caused the error with the mention of Baković in the first place, or why it’s now gone away.

Posted by: Greg Egan on November 18, 2008 1:22 AM | Permalink

Re: TeXnical Issues

Tonight I am not getting any error messages either. I had this problem in both Safari and Omniweb, and what they have in common is WebKit, so I’m guessing that the update to Safari included an update to WebKit, and that’s where the problem was.

Firefox 3.0.4 also seems to have a problem with this, displaying this character with something that looks like a fancy calligraphic typeface that is totally inconsistent with the surrounding text.

Posted by: Richard on November 18, 2008 3:16 AM | Permalink

Re: TeXnical Issues

Richard wrote:

Firefox 3.0.4 also seems to have a problem with this, displaying this character with something that looks like a fancy calligraphic typeface that is totally inconsistent with the surrounding text.

FWIW, I don’t find that for Firefox 3.0.4, under Mac OS 10.5.5. I get identical behaviour in Firefox and Safari, i.e. they don’t recognise &cacute; but they both correctly render &#x107; as ć without any font quirks (or errors, now).

Posted by: Greg Egan on November 18, 2008 3:39 AM | Permalink

Re: TeXnical Issues

In Firefox 3.0.4 I actually get the “c” with the acute accent. The problem is that it’s displayed with an odd typeface.

Posted by: Richard on November 18, 2008 4:06 AM | Permalink

Re: TeXnical Issues

This might be due to a difference in the fonts we have installed.

Posted by: Greg Egan on November 18, 2008 4:21 AM | Permalink

Re: TeXnical Issues

Greg Egan wrote:

I tried sticking “&cacute;” in an HTML file and looking at it with Safari 3.2, which displayed it as “&cacute;” rather than “ć”, but gave no error message. So I’m not sure what caused the error with the mention of Baković in the first place, or why it’s now gone away.

It depends on the declared document type. The browser is only supposed to reject errors if the page is declared to be a strict XHTML document. For example, the following document produces an error on Safari 3.2:

<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/ xhtml1/DTD/xhtml1-strict.dtd">

<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en">

<head>
<title>Test page</title>
</head>

<body>
<h1>Test page</h1>
<p>Here is a &amp;cacute; &mdash; &cacute;</p>
</body>
</html>

The pages on this site use a strict XHTML+MathML document type.

Robin

Posted by: Robin Houston on November 18, 2008 10:20 AM | Permalink

Re: TeXnical Issues

For example, the following document produces an error on Safari 3.2:

I pasted it into a file … no error! I tried substituting the DTD from the Café, and still no error.

I’ve never seen Safari give a list of errors at the top of a displayed page before (or since), the way it did for a couple of days over the &cacute thing on this site; usually any errors were tucked away discretely under the Activities window, with a count in the status bar below the page itself. I guess that behaviour was new to Safari 3.2, but it’s weird that I can’t get it to occur again.

And even weirder that you get it from that file, but I don’t!

Posted by: Greg Egan on November 18, 2008 11:16 AM | Permalink

Re: TeXnical Issues

For some reason, for the last couple of days I have been unable to view the blog in Internet Explorer: I just get a load on XML code. Does anyone have any idea why?

Posted by: Simon Wadsley on November 25, 2008 4:12 PM | Permalink

Re: TeXnical Issues

I have the same problem.

It is breaking just after the post with the c-acute, so perhaps whatever was done to fix it in Safari has broken it in IE7. It seems to be to do with a “default namespace declaration attribute” construct in the DTD which is not supported in MSXML.

It only breaks on the main page so perhaps it will work again after some posts roll off.

Posted by: PhilG on November 28, 2008 9:44 PM | Permalink

LaTeX sandbox

Some text inside the equations:

(1)minw hp h+maxw rp r+ave(w lp l) \mathop{min} w_h p_h + \mathop{max} w_r p_r + \mathop{ave}(w_l p_l)

Another equation:

(2)𝔛𝒢(x)= e x 2/2dx=2π \mathfrak{X} \Rightarrow \mathcal{G} (\stackrel{\rightarrow}{x}) = \int_{-\infty}^\infty e^{-x^2/2} \mathrm{d}x = \sqrt{2\pi}

The Dirac equation:

(3)The Dirac equation is (iD+m)ψ=0 \text{The Dirac equation is } \; \; (i\slash{D}+m)\psi =0

Some Clifford algebras:

(4)𝒞 j ,,,,(2),(4),(2)(2) \mathcal{C}\ell_{j}^- \quad \mathbb{R}, \mathbb{C}, \mathbb{H}, \mathbb{H}\oplus\mathbb{H}, \mathbb{H}(2), \mathbb{C}(4), \mathbb{H}(2)\oplus\mathbb{H}(2)

where the generators of 𝒞 j ±\mathcal{C}\ell_{j}^\pm satisfy γ iγ j+γ jγ i=±2δ ij \gamma_i\gamma_j +\gamma_j \gamma_i =\pm 2\delta_{i j} and 𝒞 n+8 ±=𝒞 n ±(16)\mathcal{C}\ell_{n+8}^\pm = \mathcal{C}\ell_n^\pm \otimes \mathbb{R}(16) is not equal to zero.

(5)lim n k=1 n1k 2=π 26 \lim_{n \to \infty} \sum_{k=1}^n \frac{1}{k^2} = \frac{\pi^2}{6}
Posted by: Bruce Bartlett on November 28, 2008 2:00 PM | Permalink

Re: LaTeX sandbox

There are a few symbols there that aren’t rendering properly for me in the most recent version of Firefox: the first and third symbols of equation (2), and what is presumably a slashed delta in equation (3), and the symbol preceding the l j ±l_j ^{\pm}s. (How do you get that curly ll?)

Posted by: Jamie Vicary on November 28, 2008 2:44 PM | Permalink

Re: LaTeX sandbox

It ought to be $\ell$: \ell
Posted by: some guy on the street on November 28, 2008 10:41 PM | Permalink

Re: LaTeX sandbox

Jamie:

Except for the slashed ‘D’ in (3), these are characters in funky fonts. Possibly you need to install more MathML fonts? (Do you get a message to this effect when you load the page?)

Posted by: Toby Bartels on November 28, 2008 11:04 PM | Permalink

Re: LaTeX sandbox

It might be helpful to show you a picture of what my browser does; where would I send it?

Posted by: some guy on the street on November 28, 2008 10:31 PM | Permalink

Themes and fonts

Inspired by the nLab wiki, I’ve been mucking about with math fonts, CSS, themes, etc. I’ve included some screenshots below of font and theme tweaks. I would appreciate comments from Jacques and other experts since I am a novice at this game and I think I may have some things a bit mixed up.

Basically, from what I understand there’s good news and there’s bad news. The good news is that it is possible to get maths to look quite good in a browser environment. Even just about as good as standard LaTeX output in a pdf viewer (maybe better in some ways).

The bad news is that I’ve found that support for math is not a priority at all for the people who run the web. Although great in many ways, in my humble opinion Unicode has run rough-shod over mathematics… read this technical report. For instance, in the early days the powers that be refused to even believe that HH and \mathcal{H} deserved to have different Unicode variants. Anyhow.

Here are some tips and tricks I learned. I use Firefox 3 (but these tweaks should work with all versions of Firefox).

Theme

In Firefox, there is a really cool new add-on called Stylish which allows users to code their own site-specific CSS for websites, and thus achieve “themes” with different fonts, colours, sizes, etc.

Here is a theme for the n-category cafe I hacked together:

?

It’s a silly light blue-white motif, the text area of the page is a bit wider, and the comments pop-up window has a much larger textarea. Most significantly it involves a font change to DejaVu Serif (more on that below).

To use it, install the Stylish add-on for Firefox, and then install the theme.

Fonts

I found that in my computer (a laptop running Firefox 3 on Windows XP), the font scheme that ends up being produced if you follow Jacques’ instructions (if you click on that orange icon on the right with “MathML” on it) does not look very good. Here is a screenshot of what I used to get:

?

There are two things at play here: the text font and the math font. On the math side, notice how the displayed equation font for “The Dirac equation” is really dim, and how the infinities on the integral are weird, and the whole font is a bit dim in general. Did you guys used to get this output too?

I believe the text font is Georgia.

Now Jacques will help me out here if I’m wrong, but after some testing with FireBug, I believe the following is going on at the level of the math font. The basic standard math fonts used above are:

STIXNonUnicode, STIXSize1, STIXGeneral, Symbol, DejaVu Sans, Cambria Math

Believe it or not, each of these fonts gets used in the math in that screenshot! For instance, the “C” in the “Cl_j” comes from DejaVu Sans, while the “l” comes from Cambria Math. The ugly writing “The Dirac equation” comes from STIXGeneral (ok it’s not so ugly by itself — it’s based on Computer Modern! — but it looks bad in that context).

Anyhow, the good news is you can change these fonts. A list of free fonts is available here; see also here. The quick and dirty way to see how a font looks like it is to “force” a change via Tools -> Options -> Content -> Default Font (but you must uncheck “Allow pages to choose their own fonts” otherwise it won’t be implemented).

That’s a good way to quickly test out a font. But amongst other problems it has the side effect of selecting that font for everything and for all your websites. What you really need to do is to change the font using CSS. The easy way to do it in Firefox is to download the theme I provided above. Be very careful: changing the “font.mathfont-family” in the about:config file has no effect in Firefox 3 — it’s a red herring. You have to do it via a CSS declaration, like this:

math{\{ font-family: DejaVu Serif, BitStream Vera Serif, Linux Libertine, Constantia, Cambria Math, STIXNonUnicode, STIXSize1, STIXGeneral, Symbol; }\}

Here are some screenshots of what’s possible. Remember, you can change the text and/or the math fonts independently. Moreover, you can specify a whole bunch of math fonts, which MathML will use when it encounters a symbol it does not know how to parse. Mostly, I’ve just included the serif fonts below for some or other silly reason. In fact, I’ve come to believe that sans serif looks the best on a screen… for instance I really like the Verdana font which is the default text font on Instiki.

One thing though: if you’re using a laptop, make sure you’ve got antialiasing going else most fonts will look crummy. If you’re running Windows XP, go here. If you’re running Vista, it’s already setup. Other operating systems probably already have this built in.

1. My favourite, DejaVu Serif (download):

?

2. Palatino Linotype:

?

3. BitStream Vera Serif (download):

?

4. Good old Knuthian TeX Computer Modern fonts (serif and sans serif), though sadly a bit dim (download)

CMU Serif:

?

CMU Serif Extra:

?

CMU Sans Serif:

?

CMU Concrete:

?

CMU Bright:

?

5. Linux Libertine O (download

?

6. Georgia:

?

7. Euclid:

?

8. STIX General (this is not actually so bad, if you mix and match it with the other fonts to handle the things it doesn’t do very well) (download):

?

9. Times New Roman:

?

Posted by: Bruce Bartlett on November 29, 2008 5:58 AM | Permalink

Re: Themes and fonts

Thanks Bruce! I’ve installed your theme, it’s a big improvement, I like my webpages bright. I am very much a serif man, too, so this definitely makes my day.

Posted by: Jamie Vicary on November 29, 2008 5:04 PM | Permalink

Re: Themes and fonts

Installed without trouble (during a mad fit of procrastination). Spiffy!

Posted by: Blake Stacey on November 29, 2008 9:22 PM | Permalink

Re: Themes and fonts

After having messed about in font-land for a few days, I’d just like to add that jsmath is very impressive. While I appreciate what MathML is all about, a scientist who browses around the jsmath website is bound to be sorely tempted by this proposal. It seems to me that itex to MathML is a system by mathematicians (i.e. Jacques and others) for mathematicians, but which is forced to go through a middle man (the bueorocracy of the WWW). jsMath cuts out the middle man.

Reading through the report by Barbara Beeton about the history behind math and Unicode made me angry. The scientific community created the world wide web, and then it gets dictated to by the infocrats who ran it? That doesn’t sound right to me. Thanks to Donald Knuth and Leslie Lamport,
we had a fully functional advanced typographic system in LaTeX, a defacto standard. Why was this standard basically ignored? At the very least it should have been given pride of place in Unicode… a specialized segment all of its own, a holy cow. We shouldn’t have had to fight for every character! And now they’re jumbled up all over the place.

That’s what makes me admire the gung-ho attitude of jsmath… it’s similar to the “can-do” attitude that Knuth had when he made TeX. Can’t get the publishers to print my book the way I want it? Fine, I’ll make a typographic system myself.

Sorry about this rant. The math system here works great; I just get annoyed that in 2008 we’re still struggling with the best system for putting math on the web.

Posted by: Bruce Bartlett on November 30, 2008 7:47 AM | Permalink

Re: Themes and fonts

I’d just like to add that jsmath is very impressive.

Let me heartily second that! Davide Cervone has mad skilz!

If you need something that works across browsers, and your pages aren’t too math-heavy, jsMath is definitely the way to go.

But it does get really slow on math-heavy pages. A tradeoff, but one that’s sometimes worth making.

Posted by: Jacques Distler on November 30, 2008 8:23 AM | Permalink | PGP Sig

Re: Themes and fonts

I’ve updated the theme slightly; it’s back to Georgia as the primary font for text, but also for math. Of course, I trust you are all editing the theme and changing the fonts and other things to your hearts content. If you find a combination which looks great, post it here! (I still like the sans serif Verdana setup on Instiki, but for some reason it doesn’t look as good here at the n-category cafe…?)

My only beef with the STIX fonts is that they’re a bit dim and and ‘dead’ looking. It looks especially bad when placed alongside a brighter font like Georgia or DejaVu Serif.

Jacques, would it be possible to correct this behaviour of the MathML system? Basically the problem is that if a user inserts text into an equation, eg. via \text{mytext} or \mathop{}, then it uses the math font to render that text, instead of the text font which has been used in the surrounding text before and after the equation. This is a bug, right?

For instance, in the default n-category cafe setup, the text font is Georgia and the math font is STIXNonUnicode. These two fonts look very different, so when a user types text into an equation, it looks pokey.

Posted by: Bruce Bartlett on November 30, 2008 1:23 AM | Permalink

Re: Themes and fonts

As a general rule, mixing different fonts for running text or for equations looks awful. For running text, you get that “Kidnapper ransom note” look and, as your example aptly demonstrates, equations look just as goofy.

You want a single font family to supply the glyphs for your equations and, ideally, a closely-related family for your running text. Unfortunately, the STIX fonts are not designed for running text (I’ve tried).

There’s another issue that affects equations: proper placement of the glyphs by the MathML renderer requires detailed knowledge of the font metrics. And, indeed, the MathML renderer in Gecko is elaborately tweaked for particular (known) mathematical fonts.

The big improvement heralded by the advent of the STIX fonts was the result of finally being able to design the rendering about a single font, which everyone would have, and which has the comprehensive glyph coverage that one needs.

I don’t have DejaVu Sans or Cambria Math installed, and my view of the same equations as in your example looks much better because all the glyphs are drawn from the STIX family.

I agree with you that STIX, being based on Computer Modern, doesn’t look as good onscreen as some other fonts which have been designed with screen viewing in mind.

Strange how some people think Computer Modern is the cat’s pajamas, and we should all be making GIF pictures of our TeX equations. No accounting for tastes …

But I disgress …

Basically the problem is that if a user inserts text into an equation, …, then it uses the math font to render that text, instead of the text font which has been used in the surrounding text before and after the equation. This is a bug, right?

One can use CSS to style <mtext> elements, in the same way you used it to style the <math> element.

I don’t know whether it looks better to match the font in <mtext> to the surrounding equation or to the font used in the running text. More experimentation is warranted.

Anyway, let me say that there are other æsthetic matters that fry my oyster more than this one. Compare

\int_{-\infty}^\infty e^{-a x^2} dx = \sqrt{\frac{\pi}{a}}

which produces e ax 2dx=πa \int_{-\infty}^\infty e^{-a x^2} dx = \sqrt{\frac{\pi}{a}} and

{\int_{-\infty}^\infty e^{-a x^2} dx = \sqrt{\frac{\pi}{a}} }

which produces e ax 2dx=πa {\int_{-\infty}^\infty e^{-a x^2} dx = \sqrt{\frac{\pi}{a}} }

Posted by: Jacques Distler on November 30, 2008 8:13 AM | Permalink | PGP Sig

Re: Themes and fonts

One can use CSS to style mtext elements, in the same way you used it to style the math element.

Great! Thanks for the tip, silly me for not realizing this.

I don’t have DejaVu Sans or Cambria Math installed, and my view of the same equations as in your example looks much better because all the glyphs are drawn from the STIX family.

Mmm… that confuses me a bit, because if I order math{font-family: STIXNonUnicode, STIXSize1} in CSS, I get this:

?

While if I order math{font-family: STIXNonUnicode, STIXSize1, STIXGeneral}, I get this:

?

Those are all the plausible combinations of STIX fonts I can think of. I’m curious — which combination is being used on your system? Can you give a screenshot of how your math is coming out? Maybe things look better on a Mac.

I agree with you that STIX, being based on Computer Modern, doesn’t look as good onscreen as some other fonts which have been designed with screen viewing in mind.

That STIXNonUnicode font looks quite nice as a text font actually, I think. It seems to me there are two text-style fonts in the STIX collection: STIXNonUnicode and STIXGeneral. The former looks more like Georgia, and the latter looks more like Computer Modern, hence unsuited for a display environment. Maybe a suitable combination can improve things a lot.

The big improvement heralded by the advent of the STIX fonts was the result of finally being able to design the rendering about a single font, which everyone would have, and which has the comprehensive glyph coverage that one needs.

Well I didn’t know that was what it was all about. This philosophy I would strongly disagree with… for me it represents one of the few advantages our MathML system here has (for now) over jsMath: we can freely change our fonts. It seems very rigid and old-school to me to suddenly stipulate that the STIX fonts are the standard for all math on the web, and that there will be a significant degradation in quality if one tries to use other fonts. That doesn’t mesh well with the idea of artistic freedom on the web. Okay, I’m happy with there being a standard, but if it means that browsers start getting lazy and building all their rendering code around STIX, that would be a big step backwards.

Posted by: Bruce Bartlett on November 30, 2008 6:46 PM | Permalink

Re: Themes and fonts

That STIXNonUnicode font looks quite nice as a text font actually…

I didn’t try that. I tried STIXGeneral. One of the (several) problems was the lack of ligature information, which normal fonts, intended for running text, possess.

This philosophy I would strongly disagree with… for me it represents one of the few advantages our MathML system here has (for now) over jsMath: we can freely change our fonts. It seems very rigid and old-school to me to suddenly stipulate that the STIX fonts are the standard for all math on the web, and that there will be a significant degradation in quality if one tries to use other fonts.

I agree.

And I’m not saying there’s no support for other fonts. I believe all the ones listed in the default font.mathfont-family are supported. I’m just saying that you can’t specify some random (‘unsupported’) font, and expect it to work well.

Can you give a screenshot of how your math is coming out? Maybe things look better on a Mac.

Here’s a screenshot. I believe all of the glyphs in the equations are from the STIX family, except for \ell, which seems to come from Arial-italic. I did not set any CSS styles for Math (beyond what’s specified in this blog’s CSS file), nor have I set the font.mathfont-family preference, not do I have DejaVu Sans or Cambria Math installed. This is the default behaviour with (just) the STIX fonts installed.

Posted by: Jacques Distler on November 30, 2008 9:16 PM | Permalink | PGP Sig

Re: Themes and fonts

Here’s a screenshot.

Wow, things really do look better on a Mac. Not just the math in particular but everything… The Georgia plain text font is also rendering better. I suspected there was some effect of this kind when I saw on the jsMath page that pc users had to download a much thicker set of Computer Modern fonts than mac users. It’s pretty funny about the \ell character coming from Arial-italic though… who would have thought that Arial italic would come to the rescue of the STIX fonts.

Posted by: Bruce Bartlett on November 30, 2008 10:23 PM | Permalink

Re: TeXnical Issues

$ \mathbf{B} $

produces

B\mathbf{B}

I see this as an italic, non-boldface B on my Mozilla Firefox browser (version 3.0.5) on my Windows machine.

$ \mathbf{BB} $

produces

BB\mathbf{BB}

I see this as a boldface BB.

$ \mathbf{B B} $

produces

BB\mathbf{B B}

I see this as an italic, non-boldface B B.

I mentioned this bug to Jacques Distler and he pointed me to this bug report.

Posted by: John Baez on February 3, 2009 1:41 AM | Permalink

Re: TeXnical Issues

Visiting the café today, I am struck that the list of recent comments in the sidebar of the main page seems strangely short. Has it been trimmed? If so, could it be un-trimmed? I use that list as my main way of seeing what the latest activity is on the blog, and it’s a shame to only be able to see the five most recently-active discussions.

Maybe it hasn’t been trimmed and my brain is just playing tricks on me, but even so, I’d like it to be a little longer!

Posted by: Jamie Vicary on February 26, 2009 11:20 AM | Permalink

Re: TeXnical Issues

Yes, I noticed that too. Visiting reasonably frequently, it’s a good way to keep up, as it’s unlikely over 8 threads will have been active, whereas 6 is quite possible.

Posted by: David Corfield on February 26, 2009 11:32 AM | Permalink

Re: TeXnical Issues

Have you looked into using the RSS feed? That works quite well for me via Google Reader.

Posted by: Eric on February 26, 2009 3:38 PM | Permalink

Re: TeXnical Issues

Yes, I thought it was getting shorter! Personally, I would like it to be even longer than it was, although I realise that that can’t go on forever.

Posted by: Toby Bartels on February 26, 2009 10:53 PM | Permalink

Re: TeXnical Issues

I’m glad you guys more-or-less agree with me!

David: I do visit quite frequently, but it’s rare that I have sufficient excess energy to reply to something; that might only happen a couple of times a week, and then I like to see all of the discussions that are less than a few days old.

Eric: I did use the RSS feed for a while, but it’s not as good as the recent comments list used to be; I liked being able to see all the authors of the recent comments presented in a compact way (not that there’s anybody whose comments I avoid, of course ;)).

Toby: I’d prefer it to be longer than it was, too! And I don’t see why it would be difficult to set things up so that clicking on the ‘Recent Comments’ heading takes you to a page listing ALL the discussions on the café, ranked by time of most recent contribution, with the same sort of mini–comment previews available when you hover over a contributor’s name with the mouse cursor.

Jacques: any chance of lengthening the list again?

Posted by: Jamie Vicary on February 27, 2009 2:02 PM | Permalink

Re: TeXnical Issues

The oldest conversation on the recent comments list is only a few hours old now! And this conversation thread here got knocked off the list after less than a day. This change has really made the cafe a less enjoyable place for me — please, please can it go back how it used to be? Or even better, like Toby suggested, could the recent comments list be even longer than it was — say, the 15 most recently active conversations?

Posted by: Jamie Vicary on March 2, 2009 10:11 PM | Permalink

Re: TeXnical Issues

I'd also be quite happy with a link to a list of most recently commented threads, or to a list of most recent comments. That would seem to require more programming, but if the motivation to shorten the list on the side is to avoid clutter on the side, then perhaps this would be preferable.

Or a link to the Atom feed like the ones you can get at the nnLab. Something I can do in the browser is all I need.

Posted by: Toby Bartels on March 2, 2009 11:40 PM | Permalink

Re: TeXnical Issues

I sent an email to Jacques about this issue, and he wrote:

I don’t know what you’re talking about. Nothing has changed.

It is (and has always been) the most recent 5 comments on the 5 most recently commented posts (total of 25 comments).

So, either something is happening that Jacques hasn’t noticed, or people are doing something like commenting on fewer different threads.

Posted by: John Baez on March 3, 2009 2:36 AM | Permalink

Re: TeXnical Issues

Judging from Jacque's phrasing (that it's always 25 comments), I don't think that the variety of posts commented upon would make a difference.

Maybe it hasn't changed after all; it just seemed shorter to me, and Jamie and David thought so too.

But maybe David could ask Jacques to make it longer?

Posted by: Toby Bartels on March 3, 2009 9:22 AM | Permalink

Re: TeXnical Issues

The Way Back Machine tells us it was eight times five. I’ll ask Jacques.

Posted by: David Corfield on March 3, 2009 10:17 AM | Permalink

Re: TeXnical Issues

We're back to 8, hurrah!

Posted by: Toby Bartels on March 3, 2009 8:14 PM | Permalink

Re: TeXnical Issues

Toby wrote:

Judging from Jacques’ phrasing (that it’s always 25 comments), I don’t think that the variety of posts commented upon would make a difference.

I guess I misinterpreted what he wrote. I was thinking “we’ll always see 25 comments, so if the comments are all about one blog entry, the list of comments will look shorter.”

But indeed it’s always five comments per entry. And now — yay! — there are eight entries, not just five.

Posted by: John Baez on March 3, 2009 9:22 PM | Permalink

Re: TeXnical Issues

Thanks everybody, and especially Jacques, for fixing this! We are back to normal, that’s great.

I think it would stil make sense for the recent comments list to be even longer, though. At the moment, the oldest conversation on the list received its last comment yesterday, so it seems likely there are ‘active’ conversations which aren’t listed.

Posted by: Jamie Vicary on March 4, 2009 10:37 AM | Permalink

Time Zone

If I'm not imagining things, the Time Zone for the timestamps on comments here seems to have changed from GMT to GMT+0100. Maybe the server moved from England to France? Or maybe it's supposed to be on British Summer Time?, even though Britain is not on Summer Time yet.

Posted by: Toby Bartels on March 16, 2009 6:18 PM | Permalink

Re: Time Zone

As of yesterday, we're back to GMT.

Clearly the effective Time Zone is North American CT+0600, an internationally fair system since nobody actually uses it.

Posted by: Toby Bartels on November 2, 2009 9:05 PM | Permalink

Re: Time Zone

What time is it now?

Posted by: Toby Bartels on March 14, 2010 7:42 PM | Permalink

Re: TeXnical Issues

The RSS comments feed for the n-category cafe seems to be down…?

Posted by: Bruce Bartlett on June 16, 2009 8:09 AM | Permalink

Technical questions on the input interface

Arnold Neumaier writes:

I have some technical questions on the blogging software. Maybe some things can be improved and/or explained.

  1. When I use the text filter “itex to MathML with parbreaks”, how do I quote a piece of text from a previous mail? Trying to copy it with the mouse produces unintelligible output.

  2. The interface provides the possibility to “Remember personal info”, but why can’t it remember the text filter used last time? I regularly forget to set it in the first preview to what I want.

  3. It would be nice if the switch between “view chronologically” and “view threaded”, at present at the bottom of the whole page, would appear at the bottom of each message.

  4. Why are the “Previous Comments and Trackbacks” repeated in each response window? It only makes navigating in the window more difficult (tiny motions have large consequences for long discussions like this one). I’d prefer to have a larger comment window.

  5. The options are visible only before the first preview, which I found a bit of a nuisance. Also, it would be nice if they (nd the name information) appeared after the command window rather than before it, since this saves scrolling in the first round.

Posted by: Toby Bartels on September 16, 2009 9:57 PM | Permalink

Re: Technical questions on the input interface

Mike Shulman replied:

When I use the text filter “itex to MathML with parbreaks”, how do I quote a piece of text from a previous mail? Trying to copy it with the mouse produces unintelligible output.

Copy it with the mouse and put a > character in front of it. This doesn’t work for math symbols, however. Several of us have been complaining about this for a while, but no one has fixed it yet.

The interface provides the possibility to “Remember personal info”, but why can’t it remember the text filter used last time?

I’ve complained about this a couple times also, but never got any answers.

Posted by: Toby Bartels on September 16, 2009 10:04 PM | Permalink

Re: Technical questions on the input interface

When I use the text filter “itex to MathML with parbreaks”, how do I quote a piece of text from a previous mail? Trying to copy it with the mouse produces unintelligible output.

Copy it with the mouse and put a > character in front of it.

No, that doesn't work with that filter; instead see the stuff about ‘blockquote’ at the FAQ above.

Or try using the ‘Markdown with itex to MathML’ filter; it’s a lot more powerful and includes this feature.

Posted by: Toby Bartels on September 16, 2009 10:14 PM | Permalink

Re: TeXnical Issues

Is there any way for us to see the original source code of the post in order to quote diagrams and equations?

Posted by: Mike Stay on September 16, 2009 11:30 PM | Permalink

Re: TeXnical Issues

Is there any way for us to see the original source code of the post in order to quote diagrams and equations?

Whoever (John, David, or Urs) started the thread can see these, but you and I can't.

There is a way to make it so that everybody else can see your source code: sign your submissions with PGP. (That's not a design choice; it's a side effect.) Only Jacques actually does this (seek examples above); it's a little inconvenient.

Posted by: Toby Bartels on September 17, 2009 12:43 AM | Permalink

Re: TeXnical Issues

That’s nice. Maybe we should all try to sign our posts; it doesn’t seem like it should be that inconvenient, and it would sure be nice sometimes to be able to copy the source of something you’re quoting.

Posted by: Mike Shulman on September 17, 2009 4:28 AM | Permalink | PGP Sig

Re: TeXnical Issues

Hmm, I guess I fail at signing things. I tried to follow the instructions here but after previewing the clearsigned text, the comment box was emptied, so it then wouldn’t let me submit. I pasted the signed version in manually, but now it can’t find the key to verify the signature.

Posted by: Mike Shulman on September 17, 2009 4:44 AM | Permalink | PGP Sig

Adding Keys

I hate BerkeleyDB.

There’s a little database of PGP-Key/Website-URL pairs. Somehow, it got corrupted. So MT couldn’t add your key.

Fixed, now.

Posted by: Jacques Distler on September 17, 2009 9:30 AM | Permalink | PGP Sig

Re: Adding Keys

Thanks Jacques! Now I can sign my posts… but it doesn’t seem to be displaying the “raw” code any more on the verification page, which was the main point of the exercise. (-:

Posted by: Mike Shulman on September 18, 2009 6:01 AM | Permalink | PGP Sig

Re: Adding Keys

… but it doesn’t seem to be displaying the “raw” code any more on the verification page,

Huh?

Posted by: Jacques Distler on September 18, 2009 6:59 AM | Permalink | PGP Sig

Re: Adding Keys

When I click on “PGP Sig” on a signed comment, on the resulting page the text box labeled ‘The “raw” OpenPGP-signed comment’ is empty.

Posted by: Mike Shulman on September 18, 2009 8:00 AM | Permalink

Re: Adding Keys

Not for me. I can see content when clicking your signature.

Posted by: Eric Forgy on September 18, 2009 4:05 PM | Permalink

Re: Adding Keys

Ok, sorry, apparently FireGPG is to blame. I’m not sure what it thought it was trying to do, but turning off its option to “Check for PGP blocks in pages” fixed the problem.

Posted by: Mike Shulman on September 18, 2009 6:46 PM | Permalink | PGP Sig

Re: TeXnical Issues

MS: Is there any way for us to see the original source code of the post in order to quote diagrams and equations?

It should not be difficult to add a link to the source beside the the link to the Permalink.

Posted by: Arnold Neumaier on September 17, 2009 3:46 PM | Permalink

Re: TeXnical Issues

Difficult or not, it will never get done until somebody does it and the only person I know of who can do that is Jacques and I’m sure he has better things to spend his time on.

In the meantime, I think we should “suggest” that, at a minimum, David, John, and Urs sign their comments :)

I never knew that we could see the raw source if the comment (and I assume “post”) is signed.

Posted by: Eric Forgy on September 17, 2009 8:50 PM | Permalink

Re: TeXnical Issues

Jacques, is the software that runs this place open source? Do you take patches?

Posted by: Mike Stay on September 18, 2009 5:38 PM | Permalink

Re: TeXnical Issues

Jacques, is the software that runs this place open source?

Currently, no. MovableType 3 is proprietary.

But, in my copious free time, I intend to port this system to Melody, which is GPL software.

Ideally, that would happen over the course of this academic year.

Do you take patches?

Yes.

Posted by: Jacques Distler on September 18, 2009 6:45 PM | Permalink | PGP Sig

Re: TeXnical Issues

Jacques, is the software that runs this place open source?

Currently, no. MovableType 3 is proprietary.

Of course, my patches, plugins, and the templates to this blog are open-source (for whatever that’s worth).

Posted by: Jacques Distler on September 18, 2009 6:59 PM | Permalink | PGP Sig

Re: TeXnical Issues

Movable Type 3 is proprietary.

I had a look at the Movable Type webpage. It looks like the source is available (a bunch of perl scripts) but that it’s not released under any kind of open source license. Is that right?

I couldn’t find a download link for 3.3 on their site; do you know one? If I can get a copy, I’ll apply your patch and see if I can get a “View original source” link working.

Posted by: Mike Stay on September 21, 2009 5:21 PM | Permalink

Re: TeXnical Issues

MT 3.x, and earlier, were proprietary, but “source-available”. Plugins and patches were, however, freely distributable.

MT 4.x has a proprietary and a GPL’d version. Melody (linked-to above) is a fork of the latter.

I couldn’t find a download link for 3.3 on their site; do you know one?

I doubt that MT 3.x is still available for download.

In any case, I suspect you’re actually interested in the OpenPGPComment plugin and the templates for this blog.

Posted by: Jacques Distler on September 21, 2009 6:35 PM | Permalink | PGP Sig

Re: TeXnical Issues

MS: Is there any way for us to see the original source code of the post in order to quote diagrams and equations?

I noticed that while mouse-copy from Firefox to my editor garbles math formulas, mouse-copy from Firefox directly to the Comment-window in which to write replies preserves the formulas.

Unfortunately, this only holds before the first preview; for some reason, the preview activity changes the Comment entry so that when applying corrections, I see some code (unicode marker?) in place of the formulas, which makes corrections sometimes a bit of a guesswork. However, the code preserves the meaning of the formulas, as can be seen by looking at the output of the next preview.

Maybe making the Comment window display the symbol instead of its code is easier to fix…

Posted by: Arnold Neumaier on September 18, 2009 6:42 PM | Permalink

Re: TeXnical Issues

AN: I noticed that while mouse-copy from Firefox to my editor garbles math formulas, mouse-copy from Firefox directly to the Comment-window in which to write replies preserves the formulas.

Unfortunately, this doesn’t hold for everything: subskrip and superscript marking is replaced by a blank….

Posted by: Arnold Neumaier on September 19, 2009 8:10 PM | Permalink

Re: TeXnical Issues

Could we set the spam-detection threshold at the cafe a bit higher? A number of times now when I’m writing several comments at once I’ve gotten the “too many comments have been submitted from you in a short time” response.

Posted by: Mike Shulman on September 18, 2009 6:02 AM | Permalink

Re: TeXnical Issues

The “Views: Print” feature in the nLab does not work properly. It prints from the discussion part only the piece directly visible in the corresponding windows.

Posted by: Arnold Neumaier on September 21, 2009 1:28 PM | Permalink

Re: TeXnical Issues

The print view works perfectly fine with CSS rules that it knows about. It has no way of knowing about the CSS styling of “query boxes” and, so, no reason to change that styling for printing.

The correct statement of your complaint is: query boxes don’t work, as one would wish them to, in the print view.

It would be quite simple, for instance, for y’all to replace

   div.query { ... }

with something more fine-grained, like

   div.query { ... }
   #revision div.query { ... }

But, again, it’s totally up to you to ensure that the CSS feature you add to the nLab work correctly in the various views.

Posted by: Jacques Distler on September 21, 2009 3:57 PM | Permalink | PGP Sig

Re: TeXnical Issues

Hi Arnold, Jacques's answer may seem strange since this was set up for the Café, where the system administration is done almost entirely by Jacques, but your question is about the Lab, where Jacques has only written the basic Instiki software that we use but not (for example) the CSS for the query boxes. So your complaint would work better on the Forum; I've just moved it there.

(Sorry that our organisation is so haphazard; it grew organically, you know.)

Posted by: Toby Bartels on September 21, 2009 11:02 PM | Permalink

Re: TeXnical Issues

Sometimes the front page doesn't update the list of Recent Comments. I think that this can happen if I reload the front page at the same time that a new comment is accepted. For example, right now it lists 25 comments to This Week’s Finds in Mathematical Physics (Week 280) and 373 to Towards a Computer-Aided System for Real Mathematics. But in fact, I just made a 374th comment to the latter.

After I make this comment, I believe that it will be fixed. (Let's see!)

Posted by: Toby Bartels on October 3, 2009 8:35 PM | Permalink

Re: TeXnical Issues

Yep!

Posted by: Toby Bartels on October 3, 2009 8:39 PM | Permalink

Internal Server Error

About 20 minutes ago, I received upon myattempted preview of an item for the Cafe the following error message: Internal Server Error

The server encountered an internal error or misconfiguration and was unable to complete your request.

Please contact the server administrator, [no address given] and inform them of the time the error occurred, and anything you might have done that may have caused the error.

More information about this error may be available in the server error log. Apache/2.3.2-dev (Unix) PHP/5.2.6 DAV/2 mod_fastcgi/2.4.6 mod_ssl/2.3.2-dev OpenSSL/0.9.7m Server at golem.ph.utexas.edu Port 80

===========

This is now the third time within the last two weeks that I get such a message; particularly annoying is that this time it was a long message that needed 90 minutes to compose, and the submission interface does not allow one to go back to the previous version.

Is it possible to recover the submitted input? If yes, I’d appreciate to get it by email (Arnold.Neumaier@univie.ac.at).

If not, I’d appreciate if the standard window that pops up when one wants to reply to a contribution could be altered to have a memory that preserves the submitted message.

In a similar vein, it is annoying that the page that pos up after a preview does not necessarily contain the same erditing content as that that was sent for preview.

It is equivalent in display but not always equivalent in ASCII (quote symbols etc. are changed to codes), with the side effect that upon posting the result without any change, the system responds with a request to preview again because of alterations made - although only the system itself made any alterations.

Thus I often (!) need another preview and another post, which doubles the already annoyingly long response times (minutes per preview, posting even longer).

Posted by: Arnold Neumaier on October 10, 2009 7:03 PM | Permalink

Re: Internal Server Error

Sometimes when that happens, you can recover your entry by hitting the “Back” button of your browser.

The traditional wisdom is to compose your post outside the edit box and paste it as a last resort. Others can give you advice on how they do things. I still use the edit box and hit the “Back” button if I have any problems. That works for me.

By the way, the Customer Service Department for this blog consists of exactly “0” people. Jacques created it, but has more important things to worry about. Caveat emptor.

Posted by: Eric Forgy on October 10, 2009 7:35 PM | Permalink

Re: Internal Server Error

Arnold, I don’t know what combination of systems you are using, but on some browsers a right click on the window will give you a pull down menu with a “Back” option, even though there’s no back arrow on the top bar.

Posted by: Jon Awbrey on October 10, 2009 7:38 PM | Permalink

Re: Internal Server Error

That’s what I always do when this problem occurs.

Posted by: John Baez on October 11, 2009 12:45 AM | Permalink

Re: Internal Server Error

JA: I don’t know what combination of systems you are using, but on some browsers a right click on the window will give you a pull down menu with a “Back” option, even though there’s no back arrow on the top bar.

I am using Firefox version 3.0.13 for Ubuntu. If I click on “Reply to this”, a new window opens that has no control features except for resizing and repositioning, and the only pull-down menu that opens on a mouseclick allows exactly the same features. Thus the history is lost.

I don’t know how to enforce that in place of a new window a new tab in the old window opens - this would allow me to go back.

MS: I edit in an external editor; editing in textareas (especially tiny ones) is a big pain. I love this firefox plugin.

Does it preserve encoded characters? If I copy with the mouse from the web page to my xemacs editor, all mathml (including apostrophes for don’t etc.) gets translated to some encoded form, and if I copy this back to the editing window and preview it, the system no longer recognizes it as mathml. Thus I then need to recode each quoted formula explicitly, which is awkward.

MS: Do you experience the same effect I do that commenting on a post with very few existing comments is much speedier than on one (like this one) that has lots? If so, then maybe we should shift our main discussion, which is getting kind of venerable, to a new post.

I just answered one with 61 posts only. previewing was done much quicker (10 seconds, while previewing in this thread takes 20). Posting on the 61 post thread still took 50 seconds, while it had sometimes been several minutes with our now 450 post thread.

Thus a shift may be useful. It would also make navigating with the sidebar easier; the latter is already very insensitive…

Posted by: Arnold Neumaier on October 11, 2009 12:06 PM | Permalink

Re: Internal Server Error

I am using Firefox version 3.0.13 for Ubuntu.

I am using Firefox version 3.0.14 for Ubuntu 8.10.

the only pull-down menu that opens on a mouseclick allows exactly the same features

After I hit ‘PREVIEW’, I can right click outside the text area and go back. I haven't tested this with a 500 error (since I also do the next bit below), but it should work the same then.

I don’t know how to enforce that in place of a new window a new tab in the old window opens

I can right-click on ‘Reply to this’ and select ‘Open Link in New Tab’.

Posted by: Toby Bartels on October 11, 2009 12:54 PM | Permalink

Re: Internal Server Error

AN: I don’t know how to enforce that in place of a new window a new tab in the old window opens

TB: I can right-click on ‘Reply to this’ and select ‘Open Link in New Tab’.

Excellent! This works; thanks!

Posted by: Arnold Neumaier on October 12, 2009 4:32 PM | Permalink

Re: Internal Server Error

If I click on “Reply to this”, a new window opens that has no control features except for resizing and repositioning, and the only pull-down menu that opens on a mouseclick allows exactly the same features.

Yah, this annoys me too. But fortunately I can open the reply windows in a new tab instead of a new window.

I don’t know how to enforce that in place of a new window a new tab in the old window opens

As I said below, and Toby repeated, right-clicking on “Reply to this” usually brings up a pop-up menu including “Open link in new tab.” Middle-click also often works, if you have a middle mouse button.

Does it preserve encoded characters?

Yes, it does. But copy-and-paste into my Emacs (and back) works too. You may have to set your Emacs to use a Unicode-compatible coding system. I don’t know how Xemacs works, but in GNU Emacs I have the following in my .emacs file:

(prefer-coding-system 'mule-utf-8)

Actually, that’s what I used to have when I was running Emacs 22. Emacs 23 seems to use Unicode by more default, making that unnecessary.

Posted by: Mike Shulman on October 12, 2009 6:10 AM | Permalink | PGP Sig

Re: Internal Server Error

MS: in GNU Emacs I have the following in my .emacs file: (prefer-coding-system ‘mule-utf-8) Actually, that’s what I used to have when I was running Emacs 22. Emacs 23 seems to use Unicode by more default, making that unnecessary.

Putting the same into .xemacs/.emacs for Xemacs21.4.21 did not help, unfortunately.

Grabbing x 2x^2\in\mathbb{R} from the previewed version with the mouse and pasting it directly into the edit window of the preview page results in

x 2∈ℝ

losing the superscript caret. Pasting it instead into Xemacs [ patched with ‘(prefer-coding-system ‘mule-utf-8) ] and then from there into the edit window results in

x 2\u2208\u211d

losing the superscript caret and the unicode interpretation.

Posted by: Arnold Neumaier on October 12, 2009 5:43 PM | Permalink

Re: Internal Server Error

Grabbing x 2x^2\in\mathbb{R} from the previewed version with the mouse and pasting it…losing the superscript caret.

I said that copy-and-paste preserves encoded characters, not that it preserves MathML. The superscript is produced by MathML markup, not by Unicode. I wouldn’t expect copy-and-paste to copy the MathML code, and you probably wouldn’t want raw MathML pasted into your edit box anyway. What you really want is to copy-and-paste the itex source that someone typed, not the MathML. If the comment you’re replying to has been PGP-signed, like this one, then you can click on “PGP Sig” next to “Reply to this” and it will show you, among other things, a textarea box with the source of that comment which you can then copy-and-paste. Unfortunately this relies on the person you’re replying to having gone to the bit of trouble to PGP-sign their comments. (In firefox/ubuntu, I use FireGPG to sign my comments.)

Posted by: Mike Shulman on October 12, 2009 8:08 PM | Permalink | PGP Sig

Re: Internal Server Error

Oh, and regarding Unicode in XEmacs, according to these web pages you may find the following helpful:

(require 'un-define)
(set-coding-priority-list '(utf-8))
(set-coding-category-system 'utf-8 'utf-8)
Posted by: Mike Shulman on October 12, 2009 8:18 PM | Permalink | PGP Sig

Re: Internal Server Error

I always open the reply windows in a new tab rather than a new window, so that I have “back” buttons and a menu and everything. (Middle-click on “reply to this” works in unix, on windows right-click and select “open in new tab”.)

Plus I edit in an external editor; editing in textareas (especially tiny ones) is a big pain. I love this firefox plugin.

In a similar vein, it is annoying that the page that pos up after a preview does not necessarily contain the same erditing content as that that was sent for preview.

This bugs me too. When I notice that I’ve copy-pasted something with quotes in it, I often try to change them to ordinary quote-marks in my first submission so this doesn’t happen. But then sometimes I forget.

Do you experience the same effect I do that commenting on a post with very few existing comments is much speedier than on one (like this one) that has lots? If so, then maybe we should shift our main discussion, which is getting kind of venerable, to a new post.

Posted by: Mike Shulman on October 11, 2009 1:16 AM | Permalink | PGP Sig

Re: Internal Server Error

Mike wrote:

Do you experience the same effect I do that commenting on a post with very few existing comments is much speedier than on one (like this one) that has lots?

It seems that I do.

If so, then maybe we should shift our main discussion, which is getting kind of venerable, to a new post.

That would be fine. And you might take that opportunity to summarize something about the different positions that you, Arnold, Todd and Toby are taking — or get them to write summaries. I suspect that most of our readers, who haven’t studied formal logic, are by now quite bewildered. Since I have studied it, I’m enjoying your discussion immensely: it’s the first time I’ve seen someone mount a sustained attack against ‘structural foundations’ that’s significantly more intelligent than “categories suck” or “ZFC was handed down from heaven”.

Posted by: John Baez on October 11, 2009 7:05 PM | Permalink

Re: TeXnical Issues

I like to use the composite symbol “:=” for defining things, but itex treats both the colon and equals signs as relations so puts space between them.

Writing “y:=f(x)” gives

(1)y:=f(x),y:=f(x),

and writing “y\colon=f(x)” gives

(2)y:=f(x).y\colon=f(x).

I want to make “:=” into a composite mathematical relation. Any ideas?

Posted by: Simon Willerton on October 11, 2009 3:42 PM | Permalink

Re: TeXnical Issues

y\coloneqq f(x) produces

yf(x) y\coloneqq f(x)

As always, the available itex commands are documented here.

Posted by: Jacques Distler on October 11, 2009 4:50 PM | Permalink | PGP Sig

Re: TeXnical Issues

Thanks, Jacques! I knew about the list, but I didn't know about this character.

Is it not possible to get iTeX to tell MathML to put no space between two relations, or is it way too much work, or has it not been tried?

Posted by: Toby Bartels on October 11, 2009 9:43 PM | Permalink

Re: TeXnical Issues

Not sure I understand what you’re asking.

Could you provide a snippet of LaTeX markup (and, ideally, the corresponding snippet of MathML) that illustrates what you want?

Posted by: Jacques Distler on October 11, 2009 11:01 PM | Permalink | PGP Sig

Re: TeXnical Issues

The snippet is y:=f(x).

TeX puts no space between the colon and the equality sign, since they both have the catcode of a relation. But iTeX produces MathmML that renders on my browser (and apparently on Simon's, and I know also on Jon Awbrey's) with a space: y:=f(x)y:=f(x). The workaround, which you gave Simon above, is to use \coloneqq instead of :=: yf(x)y\coloneqq f(x).

I assumed that the MathML produced is supposed to have a space, but maybe this is a common bug in browsers, like the one that doesn't apply some font changes to single characters. If my assumption was right, however, then either there's a bug in iTeX or a decision to do things differently in iTeX, and that's what I'm asking about.

Posted by: Toby Bartels on October 12, 2009 2:20 AM | Permalink

Re: TeXnical Issues

The snippet is y:=f(x).

That’s incorrect, even in ordinary LaTeX. In LaTeX, you would write

 \mathrel{\mathop:}=

which would be correct. But:

  1. \coloneqq is shorter.
  2. The resulting colon is not vertically-centered (aligned with the = sign), so the output looks a little odd. There is a vertically-centered colon, \vcentcolon, but that gets even more tedious to type.

Hence \coloneqq.

Depending on what LaTeX package(s) you are using, that either produces a glyph from some font (e.g. the pxfonts), or is a macro for

 \mathrel{\mathop\vcentcolon}=

In itex, since there’s a Unicode character for this, I decided to go with that. Admittedly, the situation is less than ideal, since there is no Unicode character corresponding to (say) \Eqqcolon. To implement that, I ended up composing characters: =∷\Eqqcolon.

Jon Awbrey wrote:

$x \,-\!\!\!< y$

I’m speechless.

Posted by: Jacques Distler on October 12, 2009 3:14 AM | Permalink | PGP Sig

Re: TeXnical Issues

JD: I’m speechless.

speechlessdocumentationless\mathop{speechless} \coloneqq \mathop{documentationless}

Ja Ja Boing

Posted by: Jon Awbrey on October 12, 2009 4:15 AM | Permalink

It ain’t the documentation…

That appears to be the least of your problems AB A \mathrel{&latail;} B

(N.B.: \mathrel is new. All along, you could have written AB A &latail; B which would have looked approximately correct, even if it were semantic nonsense.)

Posted by: Jacques Distler on October 12, 2009 4:03 PM | Permalink | PGP Sig

Tantum Ergo …

What can I say, that was the workaround that worked on MediaWiki, PlanetMath, and even Knol. I only learned — had to learn — the \text{…} + unicode workaround since I came here so I haven’t got around searching the database for a good replacement symbol, since unicode &#x2919; still doesn’t look quite right for the job.

It’s the overlapping contexts of MathML + Markdown + iTeX + WebTeX + Whatever that wrecked the old semantic distinctions between text and math contexts, not I …

I read the local docs that the local faqs point to. The reason why we have standards like TeX is so folks don’t have to read the morning FanClub newsletter to know what works that day.

I understand about moving platforms, and how everyone has to reinvent the wheel in local colors and racing stripes every day, but if you’re going to mock your users, then …

Maybe development would be better served if I made a table of all the things that don’t work the way the local documentation says they’re s’posed to?

When I get some time maybe …

Posted by: Jon Awbrey on October 12, 2009 5:08 PM | Permalink

Re: Tantum Ergo …

Maybe development would be better served if I made a table of all the things that don’t work the way the local documentation says they’re s’posed to?

That would, doubtless, be far more productive …

Posted by: Jacques Distler on October 12, 2009 6:32 PM | Permalink | PGP Sig

Re: Tantum Ergo …

but if you’re going to mock your users

If you meant that bit about ‘semantic nonsense’, I don't think that Jacques was trying to mock you.

If you check the PHP signature and look at Jacques's input, the first version is

A \mathrel{&latail;} B

while

A &latail; B

is the second version.

And if you look at the HTML source of this page and look at Jacques's output, the first version is

<mi>A</mi><mo lspace="thickmathspace" rspace="thickmathspace">&#x2919;</mo><mi>B</mi>

while

<mi>A</mi><mi>&#x2919;</mi><mi>B</mi>

is the second version.

The obvious difference is that \mathrel has specified the proper spacing around the symbol, but there's another difference; the first version puts the symbol in <mo>, which means that it's an operator, while the second version put it in <mi>, which means that it's a stand-alone symbol like the letters ‘A’ and ‘B’. Although it affects presentation, this is a semantic difference; the first version is right and the second version is wrong.

Posted by: Toby Bartels on October 12, 2009 9:43 PM | Permalink

Re: TeXnical Issues

That’s incorrect, even in ordinary LaTeX.

???

It's perfectly correct. It's even in the TeXbook. LaTeX changes nothing from plain TeX in how it works. I use it all the time, and it looks fine to me.

I guess that the colon isn't quite precisely centred with the equality sign, but that never bothered me. If people want to try fancy tricks or new glyphs to fix that, OK. In iTeX, I'm hapy to use \coloneqq, especially since it produces a proper Unicode character.

But regardless of which symbols you want to use, it's still true that TeX puts no space between two relations, while everybody's browser seems to render the MathML that iTeX produces with a space. (Take y==f(x) as your sample if you prefer; that has no alignment problems.) I'm wondering why the difference.

Posted by: Toby Bartels on October 12, 2009 9:15 AM | Permalink

Re: TeXnical Issues

The difference is simple. TeX is a standalone program that can take as long as it likes to produce its output, iTeX is a web-program which means that it needs to be fast. Thus TeX can have lots and lots of rules and there’s no great cost in adding a new rule, whereas iTeX has a fairly minimal set of rules and there’s a speed cost to adding new rules.

So iTeX has the rule that adjacent letters should be considered as a single entity, producing coscos from cos, because this is used quite often and saves a lot of extra commands. But when it encounters adjacent operators then it simply takes them one at a time. Thus there is no difference between := and : = as you can see: :=:= and :=: =. One could introduce new rules about when adjacent operators should be combined, but I’d guess that the list is so small it’s better simply to add new commands.

Also, it’s nothing to do with the browser. It’s all in the MathML.

Posted by: Andrew Stacey on October 12, 2009 12:06 PM | Permalink

Re: TeXnical Issues

Thanks for the answer, Andrew, but I'm not talking about spacing in the input. That makes no more difference to LaTeX (less in some cases) than to iTeX.

Maybe I'd better start over.

In plain TeX or LaTeX, if you type

y==f(x)

or

y == f(x)

or

y = = f ( x )

or anything like that, then you get output that looks more or less like

y == f(x)

If you type the same things into iTeX, then you get output that (at least when rendered on my browser, or Stephen's, or Jon's) looks more or less like

y =  = f(x)

and looks exactly like

y==f(x)y==f(x)

(on any browser looking at this).

There's a space (an extra-wide one, in fact) in the iTeX output that's not in the LaTeX output. Why is that? I suggested:

  • not possible to do with MathML,
  • too much trouble to code,
  • nobody's bothered.

Obviously more nuanced answers are possible.

Posted by: Toby Bartels on October 12, 2009 9:48 PM | Permalink

Re: TeXnical Issues

Since you clear know something about MathML, as evidenced by this comment, I’ll ask, again, what you think the MathML output, corresponding to that input should be.

To help you along, I’ll point you to the Specification for the <mo> element and, in particular, the operator dictionary from MathML2.

I will presume that you want itex2MML to handle (“correctly”, whatever that means) not just ==, but an arbitrary pair (or longer strings?) of characters corresponding to operators from the Operator Dictionary.

Now, you will probably point out that there are entries in the operator dictionary, which consist of more than one character.

In fact, "==" is one of them. If you want that entry in the operator dictionary, try \mathrel{==}, as in y==f(x) y \mathrel{==} f(x) Without the \mathrel{}, currently, what you get is a (nonsensical) sequence of two <mo> elements.

Posted by: Jacques Distler on October 13, 2009 2:48 AM | Permalink | PGP Sig

Re: TeXnical Issues

Since you clear know something about MathML, as evidenced by this comment, I’ll ask, again, what you think the MathML output, corresponding to that input should be.

I don't know as much about MathML as maybe I should. I didn't answer that question before, since I didn't know the answer and it seemed to be optional. But I'll try it now.

As you must know, but I state for others, there is a fundamental mismatch between TeX and MathML that's showing up here. When TeX sees two relations next to each other, TeX puts no space between them, while browsers don't pay any attention to the neighbours when rendering the spacing (resulting in extra-wide space in these examples instead of none). If browsers rendered neighbouring operators with no space between them (in appropriate circumstances), then the MathML currently produced by iTeX would be ‘correct’ in that it would match TeX's output, but MathML seems to have no way of telling browsers in which contexts they should do that. Also, TeX distinguishes relations from operators, while MathML does not; in fact, MathML even thinks of parentheses and the like as operators (and then rendering no space between neighbouring operators is obviously wrong). All this makes things complicated.

So what should the MathML be? There may be a better answer, but my best guess now is

<mi>y</mi><mo>=:</mo><mi>x</mi>

for y=:x (regardless of spacing). But =: is not in the dictionary, so really it needs to be

<mi>y</mi><mo form="infix" lspace="thickmathspace" rspace="thickmathspace">=:</mo><mi>x</mi>

instead; although it seems that this is supposed to render the same way be default, my browser is not doing that. One could also use

<mi>y</mi><mo rspace="0em">=</mo><mo lspace="0em">:</mo><mi>x</mi>

to more closely matching how TeX thinks about these things, although that is not semantically as good.

To do this, iTeX needs to know which operators are really operators (or relations) and which are something else, like parentheses. And even so, it still shouldn't do this for, say, y=-x, so it needs to know about TeX's distinction between relations and operators. So if we want iTeX to recreate, within MathML, the way that TeX handles these things, then it needs to learn not only which symbols should be <mi> and which should be <mo> but, more fundamentally, which symbols (and commands) are unary operators, binary operators, relations, etc, by TeX's classification. Then it can use TeX's spacing rules to assign proper lspace and rspace values to everything.

Note that it should never need to look more than one character ahead to do this. Even so, it would be a lot of work. I was under the impression that iTeX did keep track of this sort of thing, but now I see that the effects that I attributed to this are coming from the operator dictionary, not from iTeX. So if you don't think that this is worth it, then I will take your word for it; in fact, I'll be surprised if you say that it is worth it!

Well, by pointing out to me how spacing around <mo> works (with the operator dictionary), you have allowed me to figure out the answer to my question. So thank you! Also thanks for adding \mathrel to iTeX, because y\mathrel{=:}x (and y \eqqcolon x, for that example) will work out fine.

Appendix: the example above, with the three suggested MathML possibilities, in case anybody does want to try to make iTeX figure it all out:

  • y=:x
  • y=:x
  • y=:x

And here are the iTeX solutions:

  • y=:xy\mathrel{=:}x
  • yxy \eqqcolon x

(although Jacques would probably want me to adjust the vertical positioning on the first one).

Posted by: Toby Bartels on October 14, 2009 1:34 AM | Permalink

Re: TeXnical Issues

First of all, I’m glad you found this mental exercise worthwhile. It’s the kind of thought process I need to go through in implementing itex2MML. So it surely helps you to understand why itex2MML works the way it does (or, even more useful to me, to be able to point out to me places where it doesn’t work the way it should).

A couple of comments.

  • MathML’s notion of “operator” (<mo>) is rather expansive, including fences, separators and accents, as well as the things that TeX would call “operators” and “relations”.
  • But it does treat these various types of “operators” differently. And the Operator dictionary does provide defaults for their spacing, stretchiness, etc.
  • Unfortunately, those rules are not context-sensitive (ie, something like “if there are two infix operators next to each other, set the rspace on the first and the lspace on the second to zero). Probably, the reason for this is that it doesn’t make sense to have two infix operators next to each other.
  • There are a handful of 2-character entries in the operator dictionary. Arguably, itex2MML is “wrong” in not recognizing those character combinations (including your example, ==) and treating them accordingly (which, actually, doesn’t amount to anything more than converting == to <mo>==</mo> instead of <mo>=</mo><mo>=</mo>).
  • But I would be somewhat averse to extending the special treatment of multi-character “operators” beyond those mentioned in the Operator Dictionary. it just seems to me like opening up a can of worms (in terms of trying to guess what the author meant). I’m inclined to have people use explicit LaTeX commands (e.g. \eqqcolon) or specify the grouping manually (\mathrel{=∶}).
Posted by: Jacques Distler on October 14, 2009 5:50 AM | Permalink | PGP Sig

Re: TeXnical Issues

it just seems to me like opening up a can of worms (in terms of trying to guess what the author meant).

I think that we have a good default guess as to what the author meant: whatever happens in LaTeX. So if the author gives us a string of symbols or commands that are normally relations in LaTeX, then we guess that they meant them to stick together, so we combine them into a single

<mo form="infix" lspace="thickmathspace" rspace="thickmathspace">

In fact, iTeX could rely on TeX spacing rules to assign these attributes to every character, always overriding the operator table. That is probably not worth the trouble, so the current behaviour is fine.

Posted by: Toby Bartels on October 15, 2009 12:01 AM | Permalink

Re: TeXnical Issues

I suppose it’s also possible that you don’t want to have special handling for juxtapositions of arbitrary operators but, rather just those that appear as multi-character entries in the Operator Dictionary (thus, := and == but not, say, *= or =:).

It would certainly be possible to special-case that handful of entries, in itex2MML. But, while that wouldn’t be hard, it also wouldn’t buy you much. You’d have a substitute for \coloneqq but not, say, for \eqqcolon .

I’d contend that the result would not be worth the mental overhead required to remember which operators receive special treatment.

Posted by: Jacques Distler on October 13, 2009 4:07 AM | Permalink | PGP Sig

Re: TeXnical Issues

Sorry. I think I said that wrong.

My understanding is that \vcentcolon= is a synonym for \mathrel{\mathop:}=. Either one is a suitable substitute for a dedicated glyph, corresponding to \coloneqq.

But they’re both longer to type …

Posted by: Jacques Distler on October 12, 2009 3:46 AM | Permalink | PGP Sig

Re: TeXnical Issues

For instance, C.S. Peirce used a symbol for implication that was ligature of a bar - and a less than sign <\lt, as in the PlanetMath entry for Peirce’s Law.

Note. You may have to set the “view style” option to “page images” in order to see this entry.

I can usually get this with several negative spaces between the two characters, for example, $x \,-\!\!\!< y$, but negative spaces don’t seem to work in the nn-Lab.

Posted by: Jon Awbrey on October 12, 2009 2:25 AM | Permalink

Re: TeXnical Issues

Ah. Thanks Jacques. I find that slightly disturbing as I’m a bit of a perfectionist with LaTeX and I use “:=” quite a lot, yet I didn’t realise that there was a TeX command for it.

You learn something new every day.

Posted by: Simon Willerton on October 11, 2009 5:06 PM | Permalink | PGP Sig

Re: TeXnical Issues

Ah. Thanks Jacques. I find that slightly disturbing as I’m a bit of a perfectionist with LaTeX and I use “:=” quite a lot, yet I didn’t realise that there was a TeX command for it.

There isn’t. There is a command in the pxfonts package that implements it (and does so as a glyph), but the pxfonts package covers all of the standard LaTeX symbols and the AMSLaTeX symbols and a few more. As I always use this package when I can, I’ve gotten used to \coloneq, so now that I’m writing a little more iTeX, I bugged Jacques to put it in. Well, actually I made an inept attempt to put it in myself and Jacques took pity on me and did it properly.

It’s reasonably recent in iTeX. If you don’t want to follow the iTeX RSS feed to get the full list, when there’s a significant update then I put a post on the n-Forum. This particular change was announced 6 days ago

Posted by: Andrew Stacey on October 12, 2009 8:20 AM | Permalink

\coloneqq

There isn’t.

Strictly speaking, “there isn’t” much of anything in plain TeX, or vanilla-LaTeX. That’s what packages are for.

In the case at hand, there are several packages that offer a \coloneqq command. The one preferred by Instiki (i.e., the one appearing in Instiki’s TeX view) is the mathtools package – which is an adjunct to amsmath.

Posted by: Jacques Distler on October 12, 2009 4:00 PM | Permalink | PGP Sig

Pre-Established Harmony — Not !

I just use \text{:=} to get :=\text{:=}, which looks okay on Windows + Firefox 3.5.4 + STIX.

I have no idea what \mathrel{\mathop:}= parses to in other people’s moanads, Windowless, Windows-endowed, or otherwise, but it just doesn’t parse to anything from where I’m seeing it.

Folks might take a gander at the examples near the top of the nn-Lab Sandbox and discuss what they see there.

More and more often lately, I find myself resorting to \text{...} and inserting unicode from the database linked in the Sandbox.

Posted by: Jon Awbrey on October 12, 2009 2:04 PM | Permalink

Symbol Sandbox

I opened a Symbol Sandbox at the nn-Lab to study the various rendering of symbol strings.

I started a Table there to keep track of the variations, if any:

  • Column 1 is the literal code string.

  • Column 2 is the local rendering — this may show different strokes for different folks, I don’t know.

  • Column 3 is where I scribed a plaintext approximation to what I see on my screen.

  • Column 4 is where anyone who sees anything significantly other than that can record a plaintext approximation of what he or she sees.

Posted by: Jon Awbrey on October 13, 2009 5:32 AM | Permalink

Re: Symbol Sandbox

I noted a number of idiosyntactic problems in the Symbol Sandbox, some of which may be due to a bleary-eyed condition on my part that I’m going to go do something about post-post.

Posted by: Jon Awbrey on October 15, 2009 8:30 PM | Permalink

Empty Errors

I got some weird errors posting comments yesterday. When I previewed it told me I had an error, but the actual list of errors was blank. I think Dan Zaharopol had a similar problem. Did anyone else experience it? Does anyone know what might have caused it?

Posted by: Mike Shulman on October 18, 2009 2:58 AM | Permalink | PGP Sig

Re: Empty Errors

Yeah, I had a couple. By the time I'd composed an email to Jacques about it (since I couldn't comment about it here), the problem had gone away, so I never sent the email.

Posted by: Toby Bartels on October 18, 2009 6:42 AM | Permalink

Embedding video

What is the right way to embed videos?

Posted by: Simon Willerton on March 22, 2010 9:49 PM | Permalink

Re: Embedding video

What is the right way to embed videos?

I would think you should go to the page that Jacques pointed you to, choose View \to SourceCode or similar from your browser’s menu and copy-and-paste the corresponding block of code that you find there. Search the source code for “movie”, to find it.

Posted by: Urs Schreiber on March 24, 2010 2:20 PM | Permalink

Re: Embedding video

Ah, I tried that! But I only tried it in a comment and not a post! So I can put the following in this comment.

<div class="centeredfigure">
<object type="application/x-shockwave-flash" 
        data="http://www.youtube.com/v/hIkXK6rxt4c&hl=en&fs=1&rel=0&border=1" 
        width="500" height="405">
    <param name="movie" 
        value="http://www.youtube.com/v/hIkXK6rxt4c&hl=en&fs=1&rel=0&border=1" />
    <param name="allowFullScreen" value="true" />
    <param name="allowscriptaccess" value="always" />
    <p>You need the Flash Player installed.</p>
</object>
</div>

Then I get

You need the Flash Player installed.

But it works if I put it in a actual post (which I didn’t try before).

Thanks Urs.

Posted by: Simon Willerton on March 24, 2010 5:25 PM | Permalink

Remembering personal info

Is the “remember personal info” option button for comments working for other people? I told it to forget my personal info when I was reposting other people’s comments from the temporary site (so I could pretend to be them) but now I can’t get it to remember me anymore; I select “remember personal info” but then the next time I post a comment all my info is forgotten.

Posted by: Mike Shulman on March 25, 2010 3:39 AM | Permalink

Re: Remembering personal info

It's still remembering my personal info. Of course, I never told it to forget.

Posted by: Toby Bartels on March 25, 2010 4:14 AM | Permalink

Re: TeXnical Issues

This comment continues an off-topic conversation about dots above letters that we were having on a different thread.

In reply to Mike Stay’s comment:

In this aged browser (Firefox 2.0.0.10) I see the same as in the newer browser that I was using earlier today (Firefox 3.0.18), as described here.

By the way, the reason why I’m using an old browser is mostly that I’m using an old enough distribution of Linux that new versions of software often don’t install — they complain that I’ve got too old a version of GLib or dbus or something else that I entirely don’t understand. One solution would, of course, to be to upgrade my Linux installation. I could do that by (a) taking my laptop to the IT guys at work, or (b) attempting it myself. Both require some organization and energy, and so far I haven’t quite had the incentive to do it. But it’s got to the stage now where it would probably be worth it.

Posted by: Tom Leinster on March 26, 2010 12:02 AM | Permalink

WebKit

Off the post’s topic but perhaps of interest to Café folk: MathML support has now been turned on in WebKit, the Web browsing engine that runs Safari and Google Chrome.

http://webkit.org/blog/1366/announcing…mathml/

My first impression from looking at posts here is that it doesn’t fully cover MathML or, at least, doesn’t match Firefox’s rendering yet. For example, in Mike Stay’s post on Wick rotation, most necessary math formatting happens, but: it can’t find glyphs for some characters (𝒰, bold capital A); characters that shouldn’t be slanted (blackboard bold R) are; there’s too little horizontal spacing generally and too much to the left of sub/superscripts; display math is pushed to the left instead of centered; and labels to the right of expressions (starting with “unitless parametrization”) are italicized and contain no spaces, like variable names.

This would seem to be the ideal community of MathML advocates to tell Apple, Google, and co. what’s wrong or what you want at bugs.webkit.org.

I’m viewing this on a Mac, with the nightly WebKit build from nightly.webkit.org. Frequent (weekly?) WebKit builds for Windows are also available there. Dev builds of Chrome get WebKit updates on Google’s schedule, not immediately.

(Further offtopic, I have a gadget that converts LaTeX in pages to SVG (for browsers supporting it; PNGs for those that don’t) using John Forkosh’s mathTeX and pydvi2svg: mathcache.appspot.com. Adding SVG support to it was a suggestion that originally came up here. I know y’all have reasons for using MathML; only noting it since the feature was conceived in these here comment threads and y’all do math on the Web.)

(Copied over from this thread by Tom.)

Posted by: Randall Farmer on August 18, 2010 3:52 AM | Permalink

Basic worked example

For anyone who’s never figured out how to typeset math on this blog, here’s a worked example.

Background: recently I was having a conversation with Joachim Kock about how to take advantage of the mathematical typesetting capabilities of this blog. To show him how it was done, I took a long plain-text comment of his and added formatting.

There’s nothing very fancy here, but it should be enough to get you started. For more information, see the top of this page.

Here’s the source. I used the text filter “itex to MathML with parbreaks”. (When you make a comment, there’s a drop-down menu allowing you to select a text filter.) The end result is the first two-thirds or so of this comment here.

Posted by: Tom Leinster on October 8, 2010 2:16 PM | Permalink

Re: TeXnical Issues

Someone has just reported being unable to post a comment because they get the following error.

“Comment Submission Error Your comment submission failed for the following reasons:

You are not allowed to post comments.

Please correct the error in the form below, then press PREVIEW to preview your comment.”

Does this mean that they or their ip address is blacklisted or something?

Posted by: Simon Willerton on November 12, 2010 12:42 PM | Permalink

Re: TeXnical Issues

I’m not sure what it means, but it used to happen to me a lot too. Eventually I needed to email Jacques Distler, and I think it’s okay for me now.

Posted by: Bruce Bartlett on November 12, 2010 3:25 PM | Permalink

Re: TeXnical Issues

I believe this happens when your IP has been blacklisted by Jacques’ system, for spamming. Jacques should be able to deal with it. But at some point it happened to me, at my home IP address in Riverside, and we were never able to fix it. Mike Stay came up with a workaround where I’d use a proxy server when posting comments.

But that was a real nuisance, so I moved to Singapore.

Posted by: John Baez on November 13, 2010 1:27 AM | Permalink

Remembering personal details

Jim Stasheff notes that recently he’s had to retype his name and email address every time he comments. I’ve had the same thing too, on two different computers (both Firefox on Linux), for about a week. Anyone else?

Posted by: Tom Leinster on December 16, 2010 1:49 PM | Permalink

Re: Remembering personal details

I’m certainly getting that on my office machine. I thought I wasn’t on my laptop, but maybe I haven’t commented from there.

Posted by: David Corfield on December 16, 2010 2:06 PM | Permalink

Re: Remembering personal details

Yes, I have the same thing too.

Posted by: Tim Silverman on December 16, 2010 2:20 PM | Permalink

Re: Remembering personal details

Me too, for much more than a week.

Looks like the poor thing’s gone senile.

Posted by: John Baez on December 16, 2010 3:12 PM | Permalink

Re: Remembering personal details

David, Tim, John, Jim, and anyone else experiencing this: can you say which browser (with version number) you’re using? It might help Jacques. Thanks.

Posted by: Tom Leinster on December 16, 2010 3:43 PM | Permalink

Re: Remembering personal details

I’ve had this problem for months.

Posted by: Mike Shulman on December 16, 2010 3:52 PM | Permalink

Re: Remembering personal details

I’ve also had the same problem for months, although I don’t think as long as Mike. This happens on both my laptop (Mac OS X 10.5) and my desktop (Ubuntu Linux, I forget which version), both running Firefox 3.6 currently. (I suspect when the problem started I was running an earlier version of Firefox.)

Posted by: Mark Meckes on December 16, 2010 4:34 PM | Permalink

Re: Remembering personal details

I’ve had the problem for more than a week certainly and I think more than two weeks, on Firefox 3.6.12 and 3.6.13 (which was installed just today) running on Windows XP.

Posted by: Todd Trimble on December 16, 2010 6:13 PM | Permalink

Re: Remembering personal details

I’m getting the problem on Firefox 3.6.11 on CentOS Linux (as well as a much older version of Firefox, 2.0.0.something, on SUSE Linux).

Posted by: Tom Leinster on December 16, 2010 8:35 PM | Permalink

Re: Remembering personal details

I’ve had the problem so long I stopped bothering to tick the remember me box every time I posted a comment. I can’t remember when it started, but it’s been a long time. This happens on both my work computer (a Citrix windows terminal, running Firefox 3.5.16) and at home (Mac OS X 10.5. latest version of Firefox).

Posted by: David Roberts on December 16, 2010 11:40 PM | Permalink

Links not working

Over the last day or two, I’ve noticed that links don’t work properly in the list of “Recent Comments” on the front page.

When I hover my mouse over some of them, the blue text doesn’t turn white (as it normally would), no popup box appears showing the beginning of the comment (as it normally would), and worst of all, clicking the link doesn’t do anything. This is true of some post titles too.

For example, right now I can’t click on the post title “F and the Shibboleth” or on Marduk’s (first) comment on “The Eventual Image, Part 2”. Maybe about half the posts and comments are affected in this way, apparently at random - at least, I can see no pattern to it.

If I want to get to the post or comment concerned, I have to use google or some other roundabout means.

Yesterday, I noticed the same thing in the list of Recent Entries, though that’s working for me now.

I’m using Firefox 3.6.24 on CentOS Linux. I’ve also experienced this problem (today and yesterday) using an older version of Firefox on SUSE Linux.

Posted by: Tom Leinster on December 25, 2011 6:21 PM | Permalink

Re: Links not working

I think this is caused by the “Sorry we’re closed” picture, though I’m not sure how.

Posted by: Tim Silverman on December 26, 2011 12:04 PM | Permalink

Re: Links not working

I’m having the same issue, with Firefox 3.6.25 on MacOS 10.5.8. I’ve also noticed that sometimes a few links seem to work for me in one magic spot, e.g. I can click on the last two letters of a link but none of the others.

Posted by: Mark Meckes on December 26, 2011 2:52 PM | Permalink

Re: Links not working

Thanks, Tim and Mark — and Tim, what magic did you use to figure that out? Whatever it was, it seems to have worked. At least, it works for me; does it work for you too?

For the record, here’s what I changed. The original source code inserting that image was:

<div class = "centeredfigure">
<img src = "http://www.doitmyselfblog.com/wp-content/uploads/2007/11/closed2.png" alt = ""/></div>

I changed it to:

<img src = "http://www.doitmyselfblog.com/wp-content/uploads/2007/11/closed2.png" alt = ""></img>

The only drawback to this change, as far as I know, is that the image is no longer centred.

Posted by: Tom Leinster on December 26, 2011 3:29 PM | Permalink

Re: Links not working

It wasn’t very magical. I used a combination of the following 3 facts:

a) Adding the picture was the unusual new event that immediately preceded the problem.
b) Pictures and other inserted objects often make a mess on scrollable web pages with various areas.
c) The boundary between working and non-working clicks seemed to coincide with the bottom of the picture.

Posted by: Tim Silverman on December 27, 2011 8:25 PM | Permalink

Re: TeXnical Issues

Elsewhere Jim wrote:

N-category cafe and n-forum are not letting me see a list of recent postings ?? but if I click on the name of the rss feed, I do get the latest nd then can scroll down through other recent postings but this is very inefficient

help!!

apologies for posting this here totally out of context but I don’t knwo where else to complain

jim

Jesse C. McKeown replied:

Do you mean the n-Cafe main page doesn’t have all the parts it usually does?

If so, that could be a connectivity issue, such that the delivery of the page gets interrupted; and there are at least four plugs that could have a connection problem between you and whatever server Jacques’ is using. There doesn’t seem to be such a trouble at my end, so your difficulty may well disappear.

Posted by: Simon Willerton on July 9, 2013 9:23 AM | Permalink

Re: TeXnical Issues

Jim, what software are you using to read the RSS feed? The notable recent event was that google reader shut down on July 1st. I don’t know if that’s at all related.

I’ve switched to using Feedly and notice that when I click on a nn-Category post in Feedly then it opens a blank page rather than sending me to the post at the Café. If I click on an Azimuth post in Feedly then it sends me correctly to the post at Azimuth. Maybe this is connected to your problem Jim.

Posted by: Simon Willerton on July 9, 2013 9:35 AM | Permalink

Re: TeXnical Issues

Testing…

Oh, the external script ie.js expects form elements with ids comment_author, comment_email, and comment_url, but the corresponding form elements are simply ided author, email, and url. How’s about that?

Posted by: Jesse C. McKeown on September 19, 2013 5:57 PM | Permalink

Re: TeXnical Issues

Over recent years, a lot of people have mentioned that the system won’t remember their personal info for comments. Jacques has done some work on this, and it should now be fixed.

Can people who have had trouble with this in the past give it another try and leave a comment here saying whether it now works for them? Thanks.

Posted by: Tom Leinster on July 20, 2014 11:18 PM | Permalink

Re: TeXnical Issues

It works for me now! Hurrah!

(I wonder how many hours I’ve spent typing in my name each time all these years…)

Posted by: John Baez on July 21, 2014 5:30 AM | Permalink

Re: TeXnical Issues

It works for me now! Hurrah!

(I wonder how many hours I’ve spent typing in my name each time all these years…)

Posted by: John Baez on July 21, 2014 5:31 AM | Permalink

Re: TeXnical Issues

However, for some reason it’s very easy now for people to double-post comments, as above. I’ve been cleaning up a lot of these, but I still don’t know what causes it. It doesn’t seem to happen every time.

Posted by: John Baez on July 21, 2014 6:01 AM | Permalink

Re: TeXnical Issues

Works for me now, too!

Posted by: Mark Meckes on July 21, 2014 8:16 AM | Permalink

Re: TeXnical Issues

Woo-hoo! It works for me too now!

Posted by: Mike Shulman on July 21, 2014 8:19 AM | Permalink

Re: TeXnical Issues

Yes, I think so.

Hurrah!

Though I’d just been using a greasemonkey script based on the observation reported above, previously…

Posted by: Jesse C. McKeown on July 21, 2014 2:32 PM | Permalink

Re: TeXnical Issues

Has there been problems with the RSS feed? I use the new aggregator Feedly to check on posts at the Cafe and yesterday just got informed of all of the posts of the last month, since the end of November. I thought it had been unusually quiet, but then I’ve been unusually busy.

Posted by: Simon Willerton on December 20, 2014 9:11 AM | Permalink

Re: TeXnical Issues

The exact same thing happened to me.

Posted by: Mike Shulman on December 20, 2014 4:48 PM | Permalink

Re: TeXnical Issues

Me too.

Posted by: Todd Trimble on December 20, 2014 8:39 PM | Permalink

LaTeX not rendering in chronological mode on Mac Chrome

I’m running Chrome 43 on Yosemite; when I click the “view chronologically” button, none of the TeX renders at all. It works fine on Firefox and on Safari, so it’s not a Mac thing. I tried Chromium and it broke somewhere between 41 (an older version I had installed) and 45 (the latest). Assuming the version numbers are the same on Chromium and Chrome, it’s either 42 or 43.

Posted by: Mike Stay on June 1, 2015 9:27 PM | Permalink

Re: TeXnical Issues

When I try to post a comment it asks me to preview my edited comment first, when I didn’t edit my comment. Then it gives me some warning, which I don’t remember what says or if it is when I try to preview or post, before it goes through. I will see if it happens, and describe it if it does, when I try to post this comment. This is using Firefox 38.0.1 on Mac OS X 10.9.2

OK, so it says, “Your comment submission failed for the following reasons:

Please preview your modified entry before posting it.&#x201D;

Again I hadn’t edited my comment.

Then when I click “preview” I get:

“The information you have entered on this page will be sent over an insecure connection and could be read by a third party.

Are you sure you want to send this information?”

Posted by: Sam C on June 3, 2015 11:15 PM | Permalink

Re: TeXnical Issues

This happens to me whenever I copy and paste text into my comment that contains weird characters like curly apostrophes and quotation marks. I think the reason the software thinks the comment has been “modified” is that it has automatically transformed those characters into HTML entities, which of course are different from the characters themselves. Pressing submit again always works for me. (Of course, it would be nice if this could actually be fixed…)

Posted by: Mike Shulman on June 4, 2015 6:34 AM | Permalink

Re: TeXnical Issues

Thank you. So the Safari difference was probably just a coincidence. Which is good to know.

Posted by: Sam C on June 4, 2015 8:11 AM | Permalink

Re: TeXnical Issues

Safari 7.0.3 test.

Regarding my previous comment let us see what happens with Safari.

Posted by: Sam C on June 4, 2015 12:21 AM | Permalink

Re: TeXnical Issues

No problems.

Posted by: Sam C on June 4, 2015 12:26 AM | Permalink

Unwanted blank lines inserted

The purpose of this comment is to link to this sequence of comments, in which several users mention seeing a spurious blank line before every line of text in which itex is used.

Here’s a screenshot of the problem.

Posted by: Tom Leinster on February 4, 2016 5:02 PM | Permalink

Re: Unwanted blank lines inserted

Actually Qiaochu and others noted a different problem:

For me this post has a vertical line after each bit of LaTeX. Gets a bit hard to read.

Nobody showed me a picture yet, but I think they’re seeing a vertical (black?) line, not the horizontal blank line you’re seeing.

Posted by: John Baez on February 4, 2016 10:34 PM | Permalink

Re: Unwanted blank lines inserted

Actually they are reporting a different problem: they’re seeing a vertical black line after each little piece of LaTeX, not a horizontal blank line after each line containing LaTeX.

Somehow I’m mercifully immune to both problems.

Posted by: John Baez on February 4, 2016 10:37 PM | Permalink

Re: Unwanted blank lines inserted

I see; thanks. It’s easy to imagine how what one person would call a vertical space is what another would call a horizontal space, so I assumed it was the same thing.

Posted by: Tom Leinster on February 5, 2016 1:32 AM | Permalink

Re: Unwanted blank lines inserted

What browsers are people having this problem with? I have the empty space with Firefox on my mac, but not with Safari on my mac.

Posted by: Simon Willerton on February 8, 2016 11:36 PM | Permalink

Re: Unwanted blank lines inserted

Simon: Tobias and I are both having the blank line problem on Firefox 44.0 on one version or another of Ubuntu. I don’t have the problem on an up-to-date version of Tor (5.5.1) on the same operating system. This version of Tor is based on Firefox 38.6.0.

Posted by: Tom Leinster on February 9, 2016 2:13 AM | Permalink

Re: Unwanted blank lines inserted

I’m not seeing either problem on Chrome 48.0.2564.97 on OS X 10.10.5, though I did have the vertical black line problem temporarily.

Posted by: David Roberts on February 9, 2016 4:30 AM | Permalink

Re: Unwanted blank lines inserted

Jesse McKeown, who is currently having trouble posting comments here, writes the following about “Unwanted blank lines inserted”:

I have been seeing this in firefox 44.0​.1​ (linux 3.?? amd64); the offending element seems to be unstyled mathml; perhaps gecko’s mathml rendering itself is funny?

Posted by: John Baez on February 9, 2016 6:27 AM | Permalink

Re: Unwanted blank lines inserted

Jesse C. McKeown writes:

After digging through the mozilla bugzilla, it seems Fred Wang has known about this general issue for some months, and the trouble comes from fixing another problem related to different fonts’ reported size parameters; he offers this illustration; for example, on my own computer, “DejaVu Serif” and “DejaVu Sans” seem the best-tuned.

Posted by: John Baez on February 29, 2016 7:53 PM | Permalink

Re: Unwanted blank lines inserted

It seems that STIX math font was causing the problem in Firefox on my Mac. I followed the instructions here. I now have Latin modern math as my default maths font and the problem has gone away.

Posted by: Simon Willerton on March 3, 2016 10:31 AM | Permalink

Re: Unwanted blank lines inserted

Thanks for the suggestions. However, I still get the extra blank lines. (FF, Ubuntu 15.10).

Posted by: Bas Spitters on March 4, 2016 7:13 PM | Permalink

Re: TeXnical Issues

From Google Group “MathJax Users”: Extra line in MathJax 2.2

We are using MathJax 2.2 in one of our installation.

Recently we started noting a vertical line at the end of each equation …

This is MathJax issue 1300 and was fixed in MathJax v2.6.

([HTML-CSS] [Chrome 48] visual artifacts due to Chrome now rounding up [was: rendering issue in chrome canary builds] #1300)

Posted by: RodMcGuire on February 9, 2016 1:27 PM | Permalink

Re: TeXnical Issues

Yay:

\mathcal{}: 𝒜ℬ𝒞𝒟ℰℱ𝒢ℋℐ𝒥𝒦ℒℳ𝒩𝒪𝒫𝒬ℛ𝒮𝒯𝒰𝒱𝒲𝒳𝒴𝒵 \mathscr{}: 𝒜ℬ𝒞𝒟ℰℱ𝒢ℋℐ𝒥𝒦ℒℳ𝒩𝒪𝒫𝒬ℛ𝒮𝒯𝒰𝒱𝒲𝒳𝒴𝒵 \begin{split} \backslash\mathtt{mathcal}\{\}:&\,\mathcal{ABCDEFGHIJKLMNOPQRSTUVWXYZ}\\ \backslash\mathtt{mathscr}\{\}:&\,\mathscr{ABCDEFGHIJKLMNOPQRSTUVWXYZ} \end{split}

Posted by: Jacques Distler on December 13, 2016 8:27 PM | Permalink | PGP Sig

Re: TeXnical Issues

Umm, you should have linked your Dec 4 blog post MathML Update where you announce that your software now uses the STIX Two fonts for native MathML and that a latest version of the Safari browser now uses native MathML.

In the Chrome browser (which uses the MathJax kludge) both examples are displayed in the same font rather than being distinguished between Chancery and Roundhand. This is because MathJax uses its own font and currently can’t simply be told to use STIX Two.

Posted by: RodMcGuire on December 14, 2016 2:32 PM | Permalink

Re: TeXnical Issues

In the Chrome browser (which uses the MathJax kludge) both examples are displayed in the same font rather than being distinguished between Chancery and Roundhand.

Another reason to vote for Chrome to re-enable MathML.

Posted by: Jacques Distler on December 14, 2016 3:04 PM | Permalink | PGP Sig

Re: TeXnical Issues

And by ‘vote’ I guess you mean click the star next to the issue number, after being signed in with a Google account?

Posted by: David Roberts on December 16, 2016 6:24 AM | Permalink

Re: TeXnical Issues

The search function in the right-hand sidebar appears to be broken. Regardless of my search term, I get the following error message:

Got an error: Undefined subroutine CGI::Fast::multi_param at lib/MT/App/Search.pm line 68.

Posted by: Blake Stacey on December 27, 2016 9:20 PM | Permalink

Re: TeXnical Issues

I just wanted to announce a couple of TeXnical improvements that you may (or may not) notice.

  1. I’ve switched the Tikz support hereabouts to use the STIX Two fonts, instead of Computer Modern. The result is a whole lot more legible (at the default resolution) and blends much better with the existing math (which also uses the STIX Two fonts).

  2. Overflow for formulæ works again. If a formula is too long to fit in the horizontal space allotted, it sprouts a scroll bar, so that you can scroll through it. This is particularly important on mobile. This feature has a sordid history in browsers (Firefox and Safari). It was broken, then it was fixed, then it broke again… My work-around uses a bit of javascript.

Posted by: Jacques Distler on May 20, 2019 6:42 AM | Permalink | PGP Sig

Re: TeXnical Issues

Thank you!

Posted by: Blake Stacey on May 20, 2019 8:42 PM | Permalink

Re: TeXnical Issues

Tim Silverman drew lots of beautiful svg pictures here on the nn-Café:

Now none of them show up! Has something changed about the way the nnCafé displays svg figures?

Posted by: John Baez on July 28, 2019 5:27 PM | Permalink

Re: TeXnical Issues

I noticed that an SVG is failing to render here (the illustration of the Fano plane). Sometimes PNGs render correctly (they work for me here) and sometimes not (I see only the alt text here, though going directly to the file works).

Posted by: Blake Stacey on October 30, 2019 7:46 PM | Permalink

Re: TeXnical Issues

The SVGs weren’t showing for me either.

This is an update on a comment I left here

After trying a site that renders in browser/OS combinations without any luck…

…I discovered the issue for me was that, to quote from Chrome, ‘Insecure content blocked’, ‘This page is trying to load scripts from unauthenticated sources’.

After unblocking the content it now works fine. Huzzah!

Posted by: samc on November 27, 2019 6:00 AM | Permalink

Re: TeXnical Issues

The SVGs weren’t showing for me either.

This is an update on a comment I left here

After trying a site that renders in browser/OS combinations without any luck…

…I discovered the issue for me was that, to quote from Chrome, ‘Insecure content blocked’, ‘This page is trying to load scripts from unauthenticated sources’.

After unblocking the content it now works fine. Huzzah!

Posted by: samc on November 27, 2019 6:02 AM | Permalink

Re: TeXnical Issues

Daniel Cicala just reported to me that he’s receiving a very strange error when trying to post comments on the n-Cafe. When he first clicks the “post comment” link, the comment box comes up already containing a bizarre string like

undefinedsetImmediate$0.8791772468841272$1undefined

If he then replaces this string with an actual comment and clicks Preview, the comment he entered is ignored and the preview comes up with the same nonsense string in the textbox and in the preview area (as if he had left it in the textbox originally).

Has anyone else experienced this? Any ideas what could be causing it? He’s tried multiple browsers and clearing his cookies.

Posted by: Mike Shulman on August 16, 2019 4:45 PM | Permalink

Re: TeXnical Issues

Hmmm.

I have seen examples of this here and here. Obviously, it’s some kind of browser issue, but it’s unclear what.

Posted by: Jacques Distler on August 16, 2019 11:14 PM | Permalink | PGP Sig

Re: TeXnical Issues

The problem does seem to be a browser issue, but not tied to a specific browser; it occurs when I use Firefox, Chromium, or Chrome. I just received a new work desktop and the problem occurs on this machine as well. Notable is that I have my browsers on my work machine and my personal laptop (where I first noticed the issue) synced. Out of curiosity, I just tried posting a comment using a private browser, and it works. I’m not sure if this info will help to diagnose the issue, but it seems relevant.

Posted by: Daniel Cicala on August 24, 2019 6:37 PM | Permalink

Re: TeXnical Issues

Notable is that I have my browsers on my work machine and my personal laptop (where I first noticed the issue) synced.

Well, that’s the first useful thing anyone has said about this. (FWIW, the same problem was also noticed here.)

What you’re seeing is a javascript error, inscrutably splatted in the <textarea> of the comment form. But it’s not from any javascript hosted here (or, in the case of the other folks mentioned, in Instiki). So where does this javascript reside?

A good working hypothesis is that it resides in your browser-syncing code. Which is why, if you use a private browser (unsynced), the error doesn’t occur.

I don’t know if you have a way of testing this (temporarily turning off browser-syncing). But at least it’s a start…

Posted by: Jacques Distler on August 25, 2019 1:37 AM | Permalink | PGP Sig

I’m glad we’re getting closer to learning about this issue. I just unsynced my browsers, but still I receive the same error. Commenting continues to work, however, in private browsing mode.

Posted by: Daniel Cicala on August 27, 2019 4:11 PM | Permalink

Debugging

Well, maybe you need to fire up the web console in your browser and find out more about the javascript error that is producing this effect.

I assume you’ve already eliminated the obvious (like a browser extension that’s disabled in private browsing mode, but enabled otherwise).

Posted by: Jacques Distler on August 27, 2019 6:09 PM | Permalink | PGP Sig

Re: Debugging

Ahhh, looking at the extensions was not obvious to me until you mentioned it. I identified the culprit; it is the extension “Grammarly”. Once that is disabled, I can post without issue while browsing in non-private mode.

Posted by: Daniel Cicala on September 4, 2019 6:22 PM | Permalink

Re: Debugging

As Spock would say … “Fascinating.”

Thanks for debugging this. I shall have to try to warn people off using that extension.

If you feel up to it, you could try filing a bug report with them. But I’m not particularly optimistic …

Posted by: Jacques Distler on September 5, 2019 7:02 AM | Permalink | PGP Sig

Re: TeXnical Issues

In the second sentence of the “MathML” info block on the top right of your front page, the second, third, and fifth links yield “404 File Not Found” errors. The sentence is “The posts were authored in itex (an extension of [[[WebTeX]]]) using the [[[itexToMML]]] Text Filter plugin for MovableType (the [[[README]]] file).” I put triple square brackets [[[]]] around the offending links. Just thought this might be worth mentioning.

Posted by: Keith Harbaugh on September 5, 2019 4:09 PM | Permalink

People’s names in the subject field

The purpose of this comment is to flag a technical discussion happening over here. We’re talking about a little problem that’s arisen recently, whereby somehow the user’s first name accidentally shows up as the subject line in comments. (E.g. the subject line of this comment would be not “People’s names in the subject field” but “Tom”.)

Posted by: Tom Leinster on November 8, 2019 12:58 AM | Permalink

Re: TeXnical Issues

This is a very useful post for people, like me, who don’t contribute to the Café that often. A suggestion: Add to the information that appears when one makes a comment a link to this page, so that infrequent commentators can visit it and refresh their memory on various issues.

Posted by: Keith Harbaugh on July 10, 2021 5:42 PM | Permalink

Re: TeXnical Issues

Has the “random past entries” widget been reconfigured to update more frequently?

Posted by: Random Past Entries on December 29, 2021 9:39 PM | Permalink

Re: TeXnical Issues

Er, that was me, trying to change the subject line instead of the name.

Posted by: Blake Stacey on December 29, 2021 9:41 PM | Permalink

Random Past Entries

It’s updated at least every time someone leaves a comment.

That has not changed …

Posted by: Jacques Distler on December 30, 2021 4:37 AM | Permalink | PGP Sig

Random Past Entries

Ah, that clarifies things; thank you. I had somehow been under the impression that it updated daily, but I was most likely watching too casually to catch more rapid changes than that.

Posted by: Blake Stacey on December 30, 2021 6:01 PM | Permalink