The n-Category Café

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.

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.

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 “$” ($) or “\[” (\[).

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.

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.

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.

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.

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.

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.

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.

|array|, eh? Thanks, Urs!

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} ?

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

being produced by

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

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.

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: áé óú. I’m sure Windows and Unix have something comparable.

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 $$\begin{aligned} A&\overset{\quad h_1\quad}{\to}&&B\\ g_1\darr&&&\darr g_2\\ C&\underset{\quad h_2 \quad}{\to}&&D \end{aligned}$$

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.

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: $M'$ * $M\prime $ should be too low: $M\prime$ * $M^{\prime }$ should be correct: $M^{\prime }$

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.

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

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).

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.]

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

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

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.

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?

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?

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

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”.

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.

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.

I’ll second this, too.

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.

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

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

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.

Nope! Lots of other problems, but not that!

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.

Howso broken?

The MathML named entity, ∘, 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:

$$A \circ B$$

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

Thanks, Simon! Should be fixed now.

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:

$\alpha_1$ $\alpha_2$ $\alpha_3$

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

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

Bruce wrote:

In fact, I’m curious to know how various $n$-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.

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.

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.

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 $n$-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 $n$-Café? I don’t know. I really wonder about that.

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.

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.

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 $l_j ^{\pm}$s. (How do you get that curly $l$?)

It might be helpful to show you a picture of what my browser does; where would I send it?

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.

Installed without trouble (during a mad fit of procrastination). Spiffy!

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.

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.

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.

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?

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 $n$Lab. Something I can do in the browser is all I need.

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.

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.

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.

What time is it now?

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.

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.

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.

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.

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.)

Yep!

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.

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.

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.

y\coloneqq f(x) produces

$$y\coloneqq f(x)$$

As always, the available itex commands are documented here.

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.

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 $n$-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.

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.

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.

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.

It's still remembering my personal info. Of course, I never told it to forget.

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.

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.

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.

Yes, I have the same thing too.

Me too, for much more than a week.

Looks like the poor thing’s gone senile.

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.

I’ve had this problem for months.

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.)

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.

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).

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).

I think this is caused by the “Sorry we’re closed” picture, though I’m not sure how.

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.

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.

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 $n$-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.

It works for me now! Hurrah!

(I wonder how many hours I’ve spent typing in my name each time all these years…)

It works for me now! Hurrah!

(I wonder how many hours I’ve spent typing in my name each time all these years…)

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.

Works for me now, too!

Woo-hoo! It works for me too now!

Yes, I think so.

Hurrah!

Though I’d just been using a greasemonkey script based on the observation reported above, previously…

The exact same thing happened to me.

Me too.

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…)

No problems.

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.

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.

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.

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.

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.

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.

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?

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.

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.

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.

Thank you!

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).

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!

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!

Hmmm.

I have seen examples of this here and here. Obviously, it’s some kind of browser issue, but it’s unclear what.

Er, that was me, trying to change the subject line instead of the name.

It’s updated at least every time someone leaves a comment.

That has not changed …

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.