7
$\begingroup$

As has been reported in two previous questions on meta (here and here), the mathoverflow update generated a few problems with the rendering related to mathjax.

Initially there was a problem with backticks, but that was mostly solved by removing them automatically from posts. However, there remains a problem (such as here, here, and here and here, among many other examples) where the rendering seems to be completely broken, yet any edit — not matter how trivial — will cause the post to fix itself and render correctly.

Can someone explain what is going on? More importantly, will there be some update which will fix this issue automatically, or will these problems have to be fixed manually?

$\endgroup$
7
$\begingroup$

I have "re-baked" the HTML generated from the existing markdown for the current revision on all posts that exist on MathOverflow. This seems to have cleared up the cases that have been reported thus far. Please let me know if you find any other issues.

$\endgroup$
  • $\begingroup$ Great. Many thanks! What would be the easiest way to report further issues? (Another answer on this thread, a comment on this answer,...?) $\endgroup$ – Andrés E. Caicedo Jun 27 '13 at 5:35
  • 1
    $\begingroup$ Another thread would probably be best since this issue is presumed "fixed" - unless of course it isn't :) $\endgroup$ – Geoff Dalgas Jun 27 '13 at 5:38
  • $\begingroup$ Very well. Again, many thanks! $\endgroup$ – Andrés E. Caicedo Jun 27 '13 at 5:38
  • 1
    $\begingroup$ @GeoffDalgas: is it possible regenerating the HTML from markdown caused this tag wiki bug? $\endgroup$ – Anton Geraschenko Jun 27 '13 at 6:53
  • 1
    $\begingroup$ @GeoffDalgas: mathoverflow.net/posts/103839/revisions still seems to have had unwanted backticks. $\endgroup$ – jeq Jul 3 '13 at 5:13
2
$\begingroup$

Caching. The rendered HTML is cached, and what's happening is that somehow the displayed HTML has decided to parse MarkDown within MathJax dollar symbols. (MathJax is parsed on the client side, but the HTML is generated server-side)

I guess the MathJax extension was not enabled during the migration.

I'm not a dev, but I believe that forcing a rerender of all posts on MO shouldn't be hard.

$\endgroup$
  • 4
    $\begingroup$ Re-rendering we can do, we need to be able to identify the problem cases is the tricky bit. We'll see how to do this, of course we won't leave it broken. $\endgroup$ – Nick Craver Jun 26 '13 at 10:17
  • 1
    $\begingroup$ @NickCraver Hmm, OK. It seems like the problem cases are limited to paired underscores/stars in mathjax, though :) $\endgroup$ – Manishearth Jun 26 '13 at 10:20
  • $\begingroup$ It looks like there has now been a re-rendering done? In any case, the fourth example problem quoted in the question 'fixed itself' without any editing. (The other three seem fixed too, but I can't tell if it's because they were recently edited.) $\endgroup$ – Francois Ziegler Jun 27 '13 at 3:51
1
$\begingroup$

I've been fixing a few of these. I put { } around the subscripts (including the single digit ones.)

Edit I will refrain from doing more until the system-wide re-rendering is done.

$\endgroup$
  • 1
    $\begingroup$ Not necessary, though. Any edit will fix these, as long as you don't introduce more issues. $\endgroup$ – Manishearth Jun 26 '13 at 20:42
  • $\begingroup$ @Manishearth, I tried that on one edit and it didn't change. The braces fixed it. $\endgroup$ – Fred Kline Jun 26 '13 at 20:45
  • 1
    $\begingroup$ @Fred: The edit has to make a change recognized by the system. It is possible that your test edit did not change the answer, and thus did not actually count as an edit. $\endgroup$ – Ricardo Andrade Jun 26 '13 at 23:17
  • $\begingroup$ @Manishearth, You are right. I tried it on two of my own questions. $\endgroup$ – Fred Kline Jun 27 '13 at 4:29
  • $\begingroup$ @RicardoAndrade, I think that is what happened (see my other comment.) $\endgroup$ – Fred Kline Jun 27 '13 at 4:30
0
$\begingroup$

Some cases which seem not to have been dealt with so far:

\[ and \]

for generating square brackets (as in the OP Inverting the Weyl Character Formula )

and

i < j

(as in revision 1 of https://mathoverflow.net/posts/82322/revisions ). About the latter syntax, I'm surprised that it has ever worked (in LaTeX, not html!), but if it did, it's probably worth searching for it. As for the former one, I would suspect it was used often enough. I'm not sure if backslash-squarebracket actually is a meaningful command in LaTeX for generating square brackets, but people have definitely used it.

$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .