Not signed in (Sign In)

Vanilla 1.1.9 is a product of Lussumo. More Information: Documentation, Community Support.

    • CommentAuthorhgr
    • CommentTimeDec 11th 2010 edited
     
    I'm using firefox on windows. I am running Windows in High-contrast black and white mode.

    Today I found that equations on MO have become invisible, even though they I have been able to read MO perfectly Ok in the past.

    So I switched off High-contrast black and white mode, and some of the equations are now visible.

    It seems the implementation of Mathajax has changed to display images instead of Mathml.

    I sometimes browse with images turned off to speed things up, but MO is unreadable now with images turned off.

    M.SE still displays things with mathml.

    For example the following question has lots of Latex but is perfectly viewable with images turned off, and still perfectly visible with High-contrast black and white mode on.

    http://math.stackexchange.com/questions/14004/convergence-of-series-involving-iterated-sin

    Tip: If you want to see the Latex source but the question or answer hasn't been edited, just turn javascript off.
  1.  
    Note: Firefox just made an update with version 3.6.13 that broke MathJax, what a number of math web sites are using to display LaTeX. MathJax has released a update that should fix the problem as system admins update their version of MathJax.
    • CommentAuthorhgr
    • CommentTimeDec 12th 2010
     
    I'm not using 3.6.13, I'm using 3.6.12. I've just checked -it's still version 12 I'm on.

    This problem doesn't seem to be with Firefox. The change must be with Mathajax.
  2.  
    I've had the same problem today (running Google Chrome 8.0.552.215 for Windows). It works fine, however, with Firefox/3.6.12.
  3.  
    Had the same problem with a displayed line which wasn't rendering correctly. I'm also using 3.6.12.
  4.  

    I updated our MathJax fonts last night (see the MathJax news page), which is surely the source of the problem.

    Does clearing your cache help? Is this widespread enough that I should revert to the old fonts, or should we be trying to push through and solve this some other way?

  5.  

    See also http://meta.math.stackexchange.com/questions/1323/slow-processing-different-look-to-math-processing-problem-with-fonts-is-it-me.

    It seems like math.SE is also running with the new fonts. Can you confirm that you have the same problem there?

  6.  

    I suspect you can solve the problem (as well as speed up math rendering) by installing the STIX fonts. Obviously we don't want people to have to install these fonts to get the site to work, so this isn't a real solution.

    • CommentAuthorWillieWong
    • CommentTimeDec 12th 2010
     

    @Anton: installing the fonts. will report back later.

    • CommentAuthorhgr
    • CommentTimeDec 12th 2010
     
    My firefox has now automatically upgraded to 3.6.13, but the same problems still occur.

    And no I'm *not* getting these problems on Math.se

    In internet explorer 8 the display is OK when it eventually renders but it is very slow.
    • CommentAuthorHarry Gindi
    • CommentTimeDec 13th 2010 edited
     

    Someone on math.SE noted that there is a MathJax fix, and that the admins here need to update, at which point...

    all will be well.

  7.  

    I think I figured out what's going on. Short story: it's my fault.

    Long story. It looked like the MathJax svn mirror hadn't been updated in about a month. I prefer to simply check out the latest stable copy rather than try to keep track of what copies of everything I have. So I decided to clone the MathJax git repository to update the fonts. This was correct, but I forgot about one crucial bit that we need to make MathJax function properly on MO. Since MathJax is hosted on a different server from MO itself, Firefox's same origin policy prevents the webfonts from loading. This can be fixed with an .htaccess file, and now it should be.

    Please let me know if you're still experiencing any problems.

  8.  

    +1 Anton!

    • CommentAuthorhgr
    • CommentTimeDec 13th 2010
     
    Everything back to normal. Thanks Anton.