@edemaine I thought that unicode chars currently don't work in math mode.

https://github.com/Khan/KaTeX/blob/4a87f38e1a9f74883723a0aebce2fb91f1464863/src/symbols.js#L722-L736
Should they be allowed in the math mode when unicodeTextInMathMode == false?

@kevinbarabash é and other accented Unicode characters currently work in math mode, and this PR breaks that (to match XeTeX behavior).

@ylemkimon Hmm, probably not... Really, they shouldn't be supported at all right now, because they're not in the fonts. But maybe we can try to predict the future where they are in the fonts. I don't think we'll ever support Åå in math mode, for example, because math mode just doesn't have that accent (in LaTeX). So maybe we should just fully drop support for them now (in math mode only)? Scratch that, now we have \mathring! (#1125)

Çç and the other characters ought to be supported (by the font) sometime soon, but is blocked on https://github.com/KaTeX/katex-fonts/issues/2 and https://github.com/KaTeX/katex-fonts/issues/25 . I tested in XeTeX, and $ÇÐÞçðþ$ renders ð for some reason but none of the others. So we could keep these in the text symbol list, but drop them from the math symbol list. I'd personally be a fan of just dropping them altogether from math support. Or we could manually somehow add math support when unicodeTextInMathMode === true, I guess. Votes?

@edemaine got it. I forgot whether we had already made that change or not. I guess not.

I think extraLatinMath can be removed as it's in unicodeSymbols.js and #1125 was merged.

I want to get #1143 into to 0.9.0 final first. Then we can start 0.10.0-alpha with this.

@kevinbarabash That's fine, though personally I'd lean toward reducing the number of backward-incompatible changes (with non-alpha/beta changes). I'll revise this PR today for consideration, but also fine to wait on it.

Personally I actually prefer the current KaTeX behavior of "just working" with Unicode. It's a shame that LaTeX can't do that without a lot of work. Maybe the default unicodeTextInMathMode should be true? Or maybe we can change it to that once we can provide a katex.sty file that does accent mapping for text characters...

I removed the special handling of Å and å. (This should really have been done in #1125. Sorry!)

I also realized (thanks to some additional tests) that ð was being handled twice -- once in the \eth definition and once in extraLatin. I split this out (we still need an "extra" definition for ð for text mode, for now).

I think the commit for the above changes should be merged before 0.9.0 final, if possible, as they are bug fixes. I can split this into a separate PR if you prefer.

This PR is also ready to go. It now checks for extraLatin symbols and rejects them if they appear in math mode when unicodeTextInMathMode is false. When we eventually remove extraLatin, we can remove this check.

Codecov Report

Merging #1117 into master will decrease coverage by 0.01%.
The diff coverage is 100%.

@@            Coverage Diff            @@
##           master   #1117      +/-   ##
=========================================
- Coverage   79.61%   79.6%   -0.02%     
=========================================
  Files          59      59              
  Lines        3876    3879       +3     
  Branches      652     653       +1     
=========================================
+ Hits         3086    3088       +2     
- Misses        656     657       +1     
  Partials      134     134

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 7de91f7...e85d540. Read the comment docs.

I think the commit for the above changes should be merged before 0.9.0 final, if possible, as they are bug fixes. I can split this into a separate PR if you prefer.

Splitting it out in a separate PR would be the easiest b/c of the other breaking changes in this diff.

@edemaine love the "next version" tag. ^_^

I've release v0.9.0 so we'll be able to merge this after review.