Pale Moon compatibility with MathJax

[wolfbeast]

Hello,

A problem was raised with mathjax on our forum, which affects math.stackexchange.com when visited with Pale Moon.

It seems Mathjax does useragent sniffing and adapts based on which "Firefox is presented in our compatibility mode slice. For various reasons we've recently bumped this compatibility version to 60.9 from 52.9 (mainly websites complaining that the browser was too old). This is now causing issues with mathjax -- it renders properly in the main questions and answers on stackexchange, but not in the comments sections. When presenting Firefox/52.9, it does render properly everywhere. This is used as a workaround for the time being but the best solution would be if you add Pale Moon to your UA sniffing detection list and adjust as-needed for our platform support.

We can also look into making adjustments on our end to improve compatibility with mathjax, but we'll have to know exactly what the difference is between your "60.9" and "52.9" operating modes of the library.

See: https://forum.palemoon.org/viewtopic.php?f=29&t=21021

[dpvc]

we'll have to know exactly what the difference is between your "60.9" and "52.9" operating modes of the library.

There is no difference in MathJax between these two versions (in fact, the only version-specific code for Firefox is for versions less than 3.5!). The HTML-CSS output jax does produce browser-specific (and sometimes version-specific) results, but the CommonHTML and SVG output jax produce the same results in all browsers. You might try using the MathJax contextual menu to switch to one of those two and see if that works better for you. The HTML-CSS output is the oldest (and most hackish) output format in MathJax.

In version 3 (currently in beta release), there is no more browser sniffing, and the HTML-CSS output has been dropped in favor of the CommonHTML format.

[wolfbeast]

I've been able to narrow this down a bit further with your information -- it may be a specific issue with stackexchange in combination with MathJax here in that it provides a different CSS display: mode for interjected Mathjax in comments there, which happens to trigger a (frame construction) bug in our code; the contextual menu's modes for MathJax didn't make a difference there.

Glad to hear you are removing all UA sniffing from the lib in the next version! 👍

I'll close this issue because it seems to be a bug on our side which we need to fix.

[dpvc]

Thanks for the additional information. Best of luck with resolving the issue.