Davide Cervone

@JyrkiLahtonen, also, with the new extension, about how big are the clumps of characters? And what are the two numbers showing in to the right of the "Refresh Preview" (on average)?

@JyrkiLahtonen, you mention that "the characters will be delayed, and later appear one at a time". I will have to experiment and see if I can reproduce that. Does all the math get typeset between each keypress? I'm trying to figure out if the canceling of MathJax isn't working for you.

@JyrkiLahtonen, do they appear as you type them, or do you type a bunch of them but the display of the characters doesn't show up right away? After you stop typing do they seem to continue to type for a while after that? (This is for without the extension.)

During the time that MathJax is doing the typesetting of the math, the browser will be trying to take up all the time it can (it is 100% computable); on a dual-processor system, you should see 50% CPU for the browser during that time. That is perfectly normal, and does not indicate a problem. After you stop typing, MathJax will take a few seconds to finish typesetting the long answer, which accounts for the half a minute delay before returning to normal. For the short answer, the processing finishes before the next keystroke, so the process is not CPU bound as with longer posts.

@JyrkiLahtonen, can you say more about what you mean by "became quite slow again"? As you type, are the keystrokes showing up in the editing box as you type them (not the preview, but where you are typing)? Or do they show up in clumps only after you pause typing?

@JyrkiLahtonen, I am unable to reproduce the problem by editing the two answers you suggest. I editing continually for three minutes, typing pretty much constantly, at various speeds, but everything worked perfectly.

@BillDubuque, I think the correct characterization is "This is a known problem with Chrome when using MathJax in MSE's previewer." :-)

I'm afraid I'm going to have to throw in the towel on this Chrome issue, and hope Chrome takes care of itself in future versions. I doubt it is worth trying to find a workaround in MathJax with nothing really to go on.

interesting. When I originally rewrote the preview code, I improved its handling of how Markdown and MathJax interacted. That did lead to inconsistency between the preview and the post results, since the markdown for the post is processed by different code. But I thought they fixed that. Perhaps that was only for math.stackexchange.com

It might be that if MathJax did things in a different order or something, it might help, but who knows. Yes, wmd could also play a role. It does a lot of string substitutions, so perhaps there is some memory leak there.

I really think it is a Chrome bug that is triggered by the intense activity involved in the MathJax typesetting, but I don't think MathJax is doing anything illegal per se.

I'm afraid that I am out of ideas about how to analyze this bug.

OK, then that doesn't indicate cache failure after all. Sorry for the false alarm.

I assume that you have checked the tabs and they really are broken? I don't use Chrome myself very much, so I'm not familiar with how it works, but if the actions in one tab (I may have said "window" before) are affecting another, then something is seriously wrong. Unless Chrome is intentionally blocking all the tabs when one goes bad since it now longer trusts that site. Is that usually what happens with an "Aw Snap" if you have more than one tab open for that site?

You mention that all the tabs show "Aw Snap". Does that happen immediately when one tab crashes, or do you have to start editing in the other ones to make that happen?

You mention "I'll try clearing the cache, but I'm not sure how that will help catching the bug". My reason for that was that if the problem in one window is affecting the results in another window, which isn't supposed to happen, that might be caused by a corruption to the cache (due to a memory management problem, for example). I was trying to determine if refreshing the cache would clear the error, which would indicate that the cached copy was in fact corrupt.