The n-Category Café

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 $R_{\mu }{}^j{}^k{}_l$ and ${}_{2}F_1$, 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.

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.

I’m not reproducing your problem, Mike. Here is $x^{*}$ ($x^{*}$) and here is the same thing as a display equation

($$x^{*}$$). Both seem to work just fine for me.

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} ?

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: áéíóú. 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

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: $M\prime$ * $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.]