6
\$\begingroup\$

In Chrome, the EE.SE LaTeX test page crashes the tab. Link below:

Test the new LaTeX markdown in this Sandbox question!

The page partially loads (at least everything near the top for me), and then I'm presented with an error message:

I'm getting a stack overflow on StackOverflow

It doesn't look like the question or answers have been recently edited, but there's also no indication how long this has been the case. Thanks to @rdtsc for bringing it up.

\$\endgroup\$
20
  • 6
    \$\begingroup\$ "STATUS_STACK_OVERFLOW".... hahaha \$\endgroup\$
    – rdtsc
    Commented Apr 12, 2021 at 14:33
  • \$\begingroup\$ The irony was not lost on me :) \$\endgroup\$
    – W5VO
    Commented Apr 12, 2021 at 14:38
  • 2
    \$\begingroup\$ Works in Firefox 87.0 but crash in Chrome Version 89.0.4389.114 (and if you've got the developer tools console open, it says "DevTools was disconnected from the page"). Firefox console says Loading failed for the <script> with source “secure.quantserve.com/quant.js”. and utils.js errors “veryverythinmathspace”, “verythinmathspace”, “thinmathspace”, “mediummathspace”, “thickmathspace”, “verythickmathspace” and “veryverythickmathspace” are deprecated values for MathML lengths and will be removed at a future date. \$\endgroup\$
    – MarkU
    Commented Apr 13, 2021 at 5:47
  • \$\begingroup\$ The gamedev and codereview mathjax sandbox meta questions are displayed correctly... seems like the issue is limited to electronics.stackexchange.com \$\endgroup\$
    – MarkU
    Commented Apr 13, 2021 at 5:56
  • \$\begingroup\$ See also which-stack-exchange-sites-use-mathjax \$\endgroup\$
    – MarkU
    Commented Apr 13, 2021 at 5:59
  • \$\begingroup\$ LaTex / mathjax is displayed correctly in existing qa e.g. electronics.stackexchange.com/questions/537091/… \$\endgroup\$
    – MarkU
    Commented Apr 13, 2021 at 6:01
  • \$\begingroup\$ I put it under review, we'll see what happens. \$\endgroup\$
    – Voltage Spike Mod
    Commented Apr 13, 2021 at 13:15
  • \$\begingroup\$ Similar behavior in the Opera browser (no go.) \$\endgroup\$
    – rdtsc
    Commented Apr 13, 2021 at 14:22
  • 2
    \$\begingroup\$ Yeah, it seems like there is something in one of the LaTeX statements on that page that can hard crash a browser.... but it's hard to figure out what that is when it's crashing the whole page. If nothing else, this is poorly-managed user input that could be used to sabotage a page. \$\endgroup\$
    – W5VO
    Commented Apr 13, 2021 at 15:48
  • \$\begingroup\$ Dunno if it's something on our end, or potentially a bug on Chrome, but I was able to repro, and will have a dev look into it as a part of bug duty. \$\endgroup\$
    – JNat StaffMod
    Commented Apr 14, 2021 at 14:49
  • \$\begingroup\$ Thanks @JNat! In the short term, is there something we can do (e.g. remove part of a question/answer) to stop the crashing on that specific page? \$\endgroup\$
    – W5VO
    Commented Apr 14, 2021 at 15:32
  • \$\begingroup\$ I honestly have no idea. If I knew which answer was causing it, we could temporarily edit it, but since the answer below points out that no specific answer seems to be causing this... ¯\_(ツ)_/¯ \$\endgroup\$
    – JNat StaffMod
    Commented Apr 14, 2021 at 15:34
  • \$\begingroup\$ Had a dev look into this, @W5VO, and it looks like there's nothing that jumps out as the cause of this. We can either just delete this sandbox post and create a new one, or alternatively selectively delete and undelete answers on Firefox and seeing how that affects the page's performance on Chrome — potentially by doing so we'd be able to identify the set of answers causing this, and fix those (or keep 'em deleted). I'm inclined to take the former option, since it's easier... thoughts? \$\endgroup\$
    – JNat StaffMod
    Commented May 6, 2021 at 13:41
  • \$\begingroup\$ @JNat Deleting the original MathJax sandbox question sounds fine from a pragmatic perspective. \$\endgroup\$
    – W5VO
    Commented May 6, 2021 at 18:02
  • \$\begingroup\$ I'm marking this as status-completed, then, @W5VO — will leave creating the new post and deleting the old one to you, unless you tell me otherwise (though I think it'd be more appropriate for you to do it). \$\endgroup\$
    – JNat StaffMod
    Commented May 11, 2021 at 9:22

1 Answer 1

Reset to default
1
\$\begingroup\$

Makes me wonder whether one of the comments or answers on that page is crashing the client-side mathjax... there are some possible candidates...

Suggest we (temporarily?) create a second sandbox (not this question because we need it for meta) and trial the likely suspects there, under chrome browser... Update: I created a draft meta question and tried each of the 22 answers, but none of them individually caused a crash. So maybe it's not an issue of bad mathjax/LaTeX code, but client-side resources being stressed?

Went digging into the timeline using firefox (because in chrome it was crashing the tab immediately after displaying the page, which is slightly different behavior than when browsing directly to the question -- now it's displaying correclty in chrome. Hm.) Anyway here's what I see under firefox:

timeline

timeline - most recently added answer 2018 Apr 24

most recently added answer 2018 Apr 24

and also this observation about a browser crash:

answer by hkBattousai

\$\endgroup\$
1
  • \$\begingroup\$ Crashes slower for me now, but still crashes Chrome in Android \$\endgroup\$
    – Scott Seidman
    Commented Apr 16, 2021 at 1:17

You must log in to answer this question.

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