status-review
1. Repeated/overlapping equations in duplicates previews (for example through review queues):
(Go to review queues, review question that has flagged duplicate, click to Close, select duplicate of...). I don't remember seeing this before the new theme (I might be wrong though, but anyway would be nice to have this fixed). This happens in Chrome 68.0.3440.106, also in IE 11.0.9600.19100. It's also worth mentioning it happens in every single case if the duplicate has mathjax in it, not just occasionally.
This is not one time occurrence, happens almost always when there is and equation.
2. Plenty of [Math Processing Error]s before and after selecting duplicate
This might be related to previous one, but symptoms are quite different, so having this separately. In previous case, there are unusually many [Math Processing Error] occurrences, before selecting and after. However before selecting it seems that it needs to have more questions in review, but still it seems quite reproducible, not a single occurrence.
Example in list of review questions:
Example after selecting question:
3. Link underline going through math expressions:
Played with CSS a little, and seems that adding display: inline-block;
does the trick, but content is then a bit taller and also forces whole link to be on one line in some cases...
EDIT 09/09/2018:
I think I've managed to fix or at least localize first issue (duplicate equations), and this also seem to decrease occurrences of second issue as well (math processing errors). Turns out that typesetting is invoked indeed twice from different parts of code. First is through MathJax.Hub.Typeset(this)
in stub.en.js
"mathjax": function() {
return this.each(function() {
"undefined" != typeof MathJax && MathJax.Hub.Typeset(this)
})
},
and second one is in full.en.js through styleCode()
:
var t = e.find(".list-originals");
return 0 == t.children().length ? (t.hide(), void 0) : (t.show(), h(t, !0), styleCode(), t.find(".item").click(function() {
For a test I have removed above call to styleCode()
from the full.en.js, and first issue went away, second one happens much less (which makes sense since it seems to depend on number of equations to be rendered). However I do not observe any new side effects after this update, but who knows why it was there, probably it was supposed to do something :). @JoeFriend , can you check it?
Also, the math processing errors, I have looked why they appear, and it happens when processInput
in mathjax.js catches exception and does return this.processError(f, a, "Input")
. The exception callstack seems to lead always to an issue that at some point, PHTMLspanElement()
in TeX-AMS_HTML-full.js returned null object which was then accessed (but it happens randomly, timing?). Also this happens only in review queues, it does not happen in questions/comments which have much more math to be rendered, it seems to suggest that something on that review queue is not yet prepared and the math is already trying to be rendered there...