@UlrikeFischer — sorry about that test, I had already updated it in the working branch but hadn't updated for a while. There are a number of tests that need updating for LuaTeX now so I'm re-running as we speak.
@WillRobertson Just found github.com/wspr/unicode-math/commit/…. Yes that's exactly what's confusing me too. But as you see the difference too the new fontloader is not the problem. Perhaps different font versions?
@HaraldHanche-Olsen I did stupid stuff because I'm lazy :-) you know, tex.ac.uk is forwarded via texfaq.net (DNS alias entry) to texfaq.org . That is mirrored to tex.org.uk for simpler html, but since Lua could not fetch https/ssl it's mirrored to latex.org.uk (non-ssl). The script fetches this and that's been put to tex.qa since that's easier to type for me.
@HaraldHanche-Olsen I still work on completing the FAQ translation to (Brazilian) Portuguese: latex.net.br/faq/brtexfaq.pdf - still in TeX, but this may go to Markdown too, and together with the German FAQ (latex.org/detexfaq.pdf, or smartphone version: latex.org/faq-klein.pdf) it can be consolidated to one workflow for the PDFs.
@JosephWright et al: unicode-math & fontspec should now mostly follow the [check-declaration] standards — there might still be a fair few edges cases but the test suites are okay. Both have updates being processed by CTAN now.
@FaheemMitha -- probably because it's winter. although it may feel like summer to us northerners, people who live there certainly experience it differently.
@WillRobertson I'm not sure if I want to use a package that is too lazy to define its own commands and then has to resort to \def\usepackage{\restoremathinternals ... to minimize the bad sideeffects ;-). (But beside this it like mathspec only sets "alphabets").
@UlrikeFischer Ah yes, I see. It's more a replacement for mathspec than unicode-math.
@UlrikeFischer Still, should be incentive for me to get back to improving the "less optimal" parts of unicode-math, though. Unfortunately my plans are always greater than my resources...
Ahhhh, and with that, it's time for bed — first proper day back at work tomorrow.
@JosephWright yes looks like this. One wonder why he didn't ask that e.g. \DeclareUnicodeMathSymbol or whatever he needed gets added to the kernel, it would make everything much easier. Sometimes I think that some people haven't really realized yet that the kernel can add new stuff ;-).
@egreg oh. What a coincidence ;-). Well that's really the point: unicode-math has __um_set_mathchar:NNnn, mathspec has \XeTeXset@mathchar, the new package redefines \set@mathchar. It would be really better to pull this back to the kernel @JosephWright @DavidCarlisle @WillRobertson
@egreg I like also \ifnum\csname g__fontspec_math_bool\endcsname=1. And that the optional argument (a sort of key-val list) shouldn't contain any spaces.
For this answer https://tex.stackexchange.com/a/390497/2288, how can I put \toprule and \bottomrule at the start and end of the glossary output, respectively?
