Thanks for the heads up hanche. I'll keep an eye on it.

I see that "Lead MathJax developer Davide Cervone will be speaking about the project at the Joint Mathematics Meetings in San Francisco, Web, Jan 15, 2010."

I'll be out of town, but if anyone happens to be at the JMM, and wants to hear him, I'd be interested to get a report.

By "commutative diagrams", which package do you mean? I tend to use xypic, and in particular xymatrix. Part of the problem here is that there are several quite good alternatives, and it's unclear which would be best/easiest to support.

I'm a little worried about the speed of MathJax. I made a page with lots of math from the MathJax preview page, and I have two copies of this page, one using MathJax (beta 1) and one using jsMath. When I load them, the MathJax version is painfully slow compared to the jsMath version.

Can others please confirm the problem? If it happens consistently across different browsers and platforms, I'd like to bring it up on the MathJax discussion forum. Maybe the MathJax team can help us figure out how to speed things up.

@Ilya: you're absolutely right. I guess I started worrying about the problem prematurely.

@Isao,

wow, pretty neat. Could you provide some samples for using it? How does one load the plugin? I couldn't see any instructions, but also didn't take the effort of going to "view source" to find out for myself.

@Isao,

it certainly looks interesting, but we'd be very conservative about deploying MathJax plugins on MathOverflow. The real difficulty is that if we migrate to Stack Exchange 2.0, we're going to likely lose direct control of the little chunk of javascript that loads MathJax, and would have to get the Stack Exchange staff to do whatever additional setup is required.

That all said, this is really great, and I'm excited to see people developing (awesome!) plugins for tools like (awesome!) MathJax. The next question is: who is going to do pgf and tikz?

I approve of the SVG rendering, although that does limit its usability. However, mixing SVG and MathML is something of a Black Art at the moment (sadly) and it's easy to get alignment out. For example, this is what I see on the main page (FF3.6.18, Linux):

Totally, no. Less than half is screwed up. But not ready for MO use...

@Anton

Thank you for using XyJax.

Have you contacted the MathJax team about the possibility of merging XyJax into the main project? I think they were interested in implementing a diagram package eventually.

I already contacted the MathJax team. I will coorporate with them to develop XyJax as MathJax extension.

Question: the demo diagram is produced by what I would call "raw xy". Does XyJax support xymatrix? I think the vast majority of uses of xy-pic are instances of xymatrix.

Of course! I have a plan to support xymatrix :-)

@Isao: Thank you for all your development effort. The new version works much better for me, all the text is visible. However, there seems to be a positioning problem, text is consistently too low. Perhaps it’s best visible in the following diagram, where I assume the intersections are supposed to match (the screenshot is from Firefox 3.5.5, but the same problem shows up in Opera 11.50):

The same screenshot also reveals that dashed lines do not work in Firefox (they do in Opera).

Finally, when I tried to open the page in Konqueror, all I got was “Unsupported Browser. Please open with Firefox/Safari/Chrome/Opera”. That’s a bit unexpected, given that the engines behind Konqueror and Safari are similar. Did the script really attempt to use some essential feature that failed, or does it disallow on principle all browsers that are not on some whitelist? If the latter, then that’s a rather bad idea.