\DocumentMetadata{pdfversion = 2.0, testphase = {phase-III,title,math}, lang = en, uncompress}
\documentclass{article}
\usepackage{unicode-math}
\begin{document}
Hello world $y = mx + c$.
\end{document}

stream
<math xmlns="http://www.w3.org/1998/Math/MathML"> <mi> 𝑦 </mi> <mo lspace="0.278em" rspace="0.278em"> = </mo> <mi> 𝑚 </mi> <mi> 𝑥 </mi> <mo lspace="0.222em" rspace="0.222em"> + </mo> <mi> 𝑐 </mi> </math>
endstream

:)

@JosephWright looks like mathml:-) (although probably @MarcelKrüger could trust the mathml renderers more and drop the lspace and rspace when they are the default values)

@DavidCarlisle yes I wondered about thst

@DavidCarlisle As we only want semantics here, I’m minded to drop anything visual

@JosephWright well no needs to be visual as well, for example if you use ngpdf (or anything else) to generate a derived html file from the pdf, the typeset math is dropped and it simply uses the mathml inline.

@JosephWright (That is more or less what we concluded as well.)

@mickep see my reply to Joseph:-)

@DavidCarlisle Yes. Same for tex->xml/mathml

texlive.net/…

@JosephWright ^^^^

@DavidCarlisle well now that luamml is on ctan we can open issues ;-) @JosephWright for the first tests I did put luamml into the latex-texmf (like tagpdf etc). Should we keep that or rely on the "external" version?

@DavidCarlisle there is no mathml. Don't you need something to enable -dev?

@UlrikeFischer yes I just remembered, I need to update at texlive.net, give me a minute...

@UlrikeFischer see your previous comment about luamml location:-)

@JosephWright @mickep @UlrikeFischer @MarcelKrüger any thoughts on my conversation with Will on the interpretation of math alphabetic characters at github.com/latex3/unicode-math/issues/… ?

easybook updated at texlive.net

@DavidCarlisle and generating formats so probably a bit more than a minute ... I will do some shopping now.

@DavidCarlisle Basically I'm just agreeing with your point: Remapping these characters seems like a bad idea, especially by default.

@DavidCarlisle I quite agree. It is simply confusing if \mitGamma comes out upright. Or is some input from the mathematical plane changes its look.

@DavidCarlisle In MathML, where can I find the easiest to parse version of the operator dictionary?

@DavidCarlisle I think you have a point.

(it is a bit confusing with all variants Margit, it, symit, ...)

@MarcelKrüger if you like xml then it's all in xml-entities/unicode.xml (that's the actual source) the mathml core version is extracted using the python script github.com/w3c/mathml-core/blob/main/tables/… the version in mathml4 is extracted by the xsl github.com/w3c/xml-entities/blob/gh-pages/opdict.xsl

@UlrikeFischer tell Will not me :-)

@UlrikeFischer For the present I'd keep a fixed one so we don't get random failures

@DavidCarlisle You are the expert - I naively assumed that from the semantics, a viewer would use a defined appearance

@JosephWright yes but if for some reason the document author adds an explicit 3cm space mid equation, they probably expect the html version of the document to have that space even if it has no obvious semantics, so if you drop all "visual" attributes they won't be happy

@UlrikeFischer no mathml :(

@DavidCarlisle it doesn't load luamml, is it installed?

@UlrikeFischer just checking now (running on teh server, I wonder if the ctan mirror was not up to date)

@UlrikeFischer that's it I'll update again from a different mirror

@JosephWright As far as I can tell there is no real way to express semantic information like: "This operator is used as a relation" without "visual" attributes, so when characters are used in a non-default setting spacing can encode semantic information.

oh no I'd missed an easybook update

tlmgr: package repository ftp.fau.de/ctan/systems/texlive/tlnet (verified)
tlmgr: saving backups to /usr/local/texlive/2024/tlpkg/backups
[1/8, ??:??/??:??] update: easybook [690k] (71999 -> 72046) ... done
[2/8, 00:04/01:59] update: harmony [166k] (15878 -> 72045) ... done
[3/8, 00:06/02:24] update: l3kernel-dev [12548k] (71856 -> 72047) ... done
[4/8, 00:09/00:13] update: magicwatermark [212k] (71998 -> 72044) ... done
[5/8, 00:11/00:16] update: simpleicons [5931k] (72000 -> 72048) ... done

@MarcelKrüger that's why we are adding <mi intent="bessel-function">J</mi> ...

@DavidCarlisle 😱

@DavidCarlisle Does that also affect default spacing in common viewers?

@MarcelKrüger no purely affects generated speech

@UlrikeFischer works now ^^^

@DavidCarlisle ;-) Could you make the log window a bit larger on texlive.net/ngpdf? Imho this is perfect for the tutorial, but one can see only around 6 lines ...

@UlrikeFischer possibly let me look...

@UlrikeFischer skype for secret reply away from ducks and dodos

@DavidCarlisle haha!

@UlrikeFischer how about if I always made the log available and just put an "open log" link alongside the open in ngpdf link?

@DavidCarlisle that would be nice too. For tagging you really need to check also the warnings.

@UlrikeFischer if you try it now you'll see there is a link to the log but you need to scroll down a bit to see it and it's currently always broken as the server hasn't copied it to where I thought it would copy

@DavidCarlisle you could put the link to the log beside the link to the PDF? Then one doesn't need to scroll.

@UlrikeFischer oh I'll make the window a couple of lines taller so scroll isn't needed, but I can't find out where the log file went:-)

who wrote this perl?

@DavidCarlisle is it not in the same place when one gets an error or use the log-directive?

@UlrikeFischer I checked with find it's not anywhere:-) basically I delete the working directory (and the users original files) as soon as possible and just move the pdf and/or the log and/or html to the public server area right at the end, I just changed it to save both log and pdf for ngpdf case but the perl and I are having a minor disagreement

@UlrikeFischer try now. Apparently you need to copy the file before deleting the directory, not after.

@DavidCarlisle nice. Can one open the log in another tab also if there is an error?

@UlrikeFischer in principle yes, let me see...

@UlrikeFischer done

@DavidCarlisle perfect!

@DavidCarlisle @JosephWright now packing for doceng youtube.com/watch?v=jqWt49o7R-k

@UlrikeFischer I've saved this version in the latexcgi git (it was just living directly on the texlive.net server as an experiment previously)