Hello guys, can I kindly ask a question?

Why sometimes Latex add space in math mode and sometimes not?

The only difference is that in the rows below I used \multicolumn

I would like to have some consistency in the table.

Does anyone know?

@Richard The chat is generally not for questions like this. Please post a question on the site, including minimal but complete source code so we know exactly what you are doing, how and can reproduce it.

I thought it was a simple question with simple solution, sorry

:)

@Richard without code it is not a simple question. I have no idea how you input that and what you mean with multicolumn. Regarding asking questions here: tex.meta.stackexchange.com/a/7292/2388

@UlrikeFischer Ok

@Richard is that a tabular with inter-column spacing? that is generally unrelated to teh math spacing around a relation

@UlrikeFischer can I blame you?

\documentclass{article}

\usepackage{fontspec}
\setmainfont{TeX Gyre Termes}
\begin{document}
 hmm
\end{document}

I deleted my luatex font cache and let it rebuild but still it uses:

This is LuaHBTeX, Version 1.18.0 (TeX Live 2024)
 restricted system commands enabled.
(./aa840.tex
LaTeX2e <2024-06-01> patch level 2
L3 programming layer <2024-08-16>
 (/usr/local/texlive/2024/texmf-dist/tex/latex/base/article.cls
Document Class: article 2024/02/08 v1.4n Standard LaTeX document class
(/usr/local/texlive/2024/texmf-dist/tex/latex/base/size10.clo))
(/usr/local/texlive/2024/texmf-dist/tex/latex/fontspec/fontspec.sty
(/usr/local/texlive/2024/texmf-dist/tex/latex/l3packages/xparse/xparse.sty

why the heck is it using the system urw times clone?

A user called "Pen and Paper" asking about chemfig? I'm so tempted to suggest to draw the molecule with pen and paper :P

@DavidCarlisle interesting. Which luaotfload version do you have? Does it also happens if you install the dev version?

luaotfload | init : Loading fontloader "fontloader-2023-12-28.lua" from kpse-res
olved path "/usr/local/texlive/2024/texmf-dist/tex/luatex/luaotfload/fontloader-
2023-12-28.lua".
Lua-only attribute luaotfload@noligature = 1
luaotfload | init : Context OpenType loader version 3.134

@UlrikeFischer ^

@UlrikeFischer you mean install luaotfload from github main branch?

get same

@DavidCarlisle github develop. But perhaps you should at first check in the luaotfload-names.lua what font data this C059-Roman.otf has.

   ["basename"]="C059-Roman.otf",
   ["familyname"]="",
   ["fontname"]="",
   ["format"]="otf",
   ["fullname"]="",
   ["fullpath"]="/usr/share/fonts/urw-base35/C059-Roman.otf",
   ["index"]=2537,
   ["italicangle"]=0x0p+0,
   ["location"]="system",
   ["pfmweight"]=400,
   ["plainname"]="C059-Roman",
   ["psname"]="",
   ["size"]=false,
   ["subfamily"]="",
   ["subfont"]=false,
   ["version"]="Version 1.00",
   ["weight"]="roman",
   ["width"]=5,
  },

oh dev ?

@DavidCarlisle yes the branch name is dev. But it is curious that the font has basically no names at all. Does it highjack only texgyretermes?

@UlrikeFischer no essentially everything, I noticed first that uniocde-math complaining that STIX TWO Math wasn't a math font

@UlrikeFischer I tried hiding that directory but it's the system Times/helvetica base clone fonts so lots of the GUI doesn't work without it

I could raise a luaotfload issue?

@DavidCarlisle yes, would be imho a good idea.

@DavidCarlisle but that must be new then, doesn't it? What changed in your system?

@UlrikeFischer not sure, it may have been a cygwin update that changed the font, perhaps..

@DavidCarlisle I get those "extended latin" with stixtwo. Your suggestion in the comment to use Termes instead of Stix, I do not understand.

@mickep see above ^^ my luatex font loading is completely broken....

@DavidCarlisle Oh, indeed, but then the comment does not make sense...

@UlrikeFischer no, all the fonts in the base35 directory all dated 2018

@mickep well if the characters are not in stix2 using a font that has them would be an option?

@DavidCarlisle Yes, but they are in opentype stix2, aren't they?

@mickep they were not in my test but then my test was using some random urw font not the font I asked for...

@mickep which started the above thread, but I'll go back and clean up the comments

@m

OK, no worries...

oh no, I initially used xelatex and got the missing chars, then tested with lualatex before things got really strange

But maybe \usepackage{stix2} uses the type1 fonts? And maybe one does not get them (automatically) there? Just guessing...

This is from the opentype font.

@DavidCarlisle you could force luaotfload to search first the texmf-tree, but I wonder what changed here ... Did something perhaps set OSFONTDIR?

@UlrikeFischer no I checked that (I had added the windows font dir but it's restored now to the default OSFONTDIR = /please/set/osfontdir/in/the/environment

@UlrikeFischer oh!! I just tried a clean run in wsl2 using linux texlive rather than cygwin and get an equvalent thing it uses /usr/share/fonts/opentype/urw-base35/C059-Roman.otf

@DavidCarlisle what happens if you blacklist the font (put C059-Roman.otf in a luaotfload-blacklist.cnf in your local directory)?

@UlrikeFischer github.com/latex3/luaotfload/issues/287

@DavidCarlisle what happens if you do \setmainfont{texgyretermes}?

@UlrikeFischer just teh file name as plain text? that seems to make no difference in the blacklist?

@DavidCarlisle yes, that is what the docu says ;-)

@UlrikeFischer that is still bad but \setmainfont{texgyretermes-regular.otf} using the filename works

@DavidCarlisle I found a version of the font, but it seems to work ok, and it does not return empty fontnames but e.g. ["fullname"]="c059roman". Lets hope that Marcel has an idea what happens here.

@UlrikeFischer Query, in the documentation for tagpdf-code you can see invisible spaces, for example on page 7 activate/mc␣(setup-key)...is the space ␣ mandatory?

@PabloGonzálezL no, that is only an artifact from my poor-man use of the macro environment.

@UlrikeFischer Ha, I see, something similar happened to me when they put \obeyline in l3doc

@UlrikeFischer Why does the use of \tagpdfparaOff not result equal to \tag_tool:n {para/tagging=false}...is it because of the internal/external structure order?

@PabloGonzálezL they should do the same. Do you have an example that shows a difference?

\DocumentMetadata{lang=en-US, pdfversion=2.0, testphase={phase-III,math,table,title,firstaid}, uncompress}% pdfstandard=ua-2,
\documentclass{article}
\usepackage[top=1cm,bottom=2cm]{geometry} % smaller picture
\usepackage{graphicx,tagpdf-debug}

\ExplSyntaxOn

\NewDocumentCommand{\Item}{som}
 {
  \seq_put_right:Nn \l_myfakelist_pictab_body_seq { \__myfakelist_pictab_item:nnn { #1 } { #2 } { #3 } }
 }
\SuspendTagging{myfakelist}
\NewDocumentEnvironment{myfakelist}{o}
 {
  \par
  \centering
  \SuspendTagging{myfakelist}

@UlrikeFischer Sorry, it's a bit long but it's the skeleton of the code I started with (the real one is a list environment)

@PabloGonzálezL ah you suspend tagging. then yes, there is a difference, tag_tool tests is tagging is active (but do not rely on it, I'm not sure if that will stay ...). Resume tagging first.

@UlrikeFischer Great, thank you very much, by the way, on page 54 you should change the tagpdf-code to say \tag_tool:n {para/tagging=false} :D

@DavidCarlisle I have a question, but it is just for explanation, the following code is valid and has always worked

\documentclass{article}
\begin{document}
\begin{enumerate}
\item \begin{minipage}[t]{3cm} \begin{enumerate} \item INERR \end{enumerate} \end{minipage}
\end{enumerate}
\end{document}

But if I put it in my list environment (tex.stackexchange.com/a/725902/7832)

\begin{document}
\begin{mylist*}
\item \begin{enumerate} \item INERR \end{enumerate}
\end{mylist*}
\end{document}

I get the error:

! Extra }, or forgotten \endgroup.
\endminipage ...pagefalse \color@endgroup \egroup
                                                  \expandafter \@iiiparbox \...

@DavidCarlisle The code I based this on (shorlst) has that limitation documented, but I don't understand why.

@PabloGonzálezL looks like you have the color safe groups messed up, but rather a lot of code to trace exactly where:-)

@DavidCarlisle Mmm, it shouldn't be, I have full control of minipage, and this only happens with the base environments provided by LaTeX, but not with different list environments, i.e. I can place a custom vertical list environment inside without any problem.

@PabloGonzálezL make an example that shows the problem (preferably with most of the key definitions removed if not needed to show the issue) , are you using \DocumentMetadata the example above does not, the answer you link to does. The list implementation is completely different if you enable tagging

@DavidCarlisle Mmm, that limitation is there from day 1, I will create a reduced MWE (without tagPDF and removing the unnecessary keys)... it is part of a long project that would already be OK if it weren't for tagPDF, now I have a kind of obsession with making it compatible (I blame @JosephWright for saying it here in the chat)

@DavidCarlisle Ohh, I can't reduce the keys because it affects the definition of \item, I could only remove the tagPDF stuff (which is the code in the original question), anyway the error is the same :(

@PabloGonzálezL in the shortlst it explodes because it somewhere does \settowidth on the text, and that doesn't like list. Do you do some measuring somewhere?

@UlrikeFischer More or less, to be able to determine the width of hbox and minipage...mmm, the strange thing is that if you create any environment using enumitem for example, it doesn't show an error (leaving aside tagPDF)

@PabloGonzálezL actually that it not the problem. You are redefining \item. Try this:

\begin{document}\let\oriitem\item
\begin{mylist*}
\item \begin{enumerate} \oriitem INERR \end{enumerate}
\end{mylist*}
\end{document}

@UlrikeFischer It's an interesting idea, I'll try it :D

@PabloGonzálezL it was more meant as demonstration that your local item definition is the problem. How you resolve that is another question.

@UlrikeFischer \def\item{\@latex@error{Lonely \string\item--perhaps a missing list environment}\@ehc} :-)

  \RenewDocumentCommand \item { o }
    {
      \tl_if_novalue:nTF {#1}
        {
          \__mylist_item_std:w
        }
        {
          \__mylist_item_std:w [#1]
        }
    }

@UlrikeFischer @DavidCarlisle Works perfectly :D...I had thought of this before, but I used \let and this doesn't work with \item (less with tagPDF)....thank you very much.

@PabloGonzálezL there is a reason we introduced \NewCommandCopy , \let only works by accident on latex commands.

@DavidCarlisle Hehe, using \let with \item is very common in the forum :D

@PabloGonzálezL I can't post every answer, so some answers are wrong

@DavidCarlisle Yes, it's true, some summer challenge updating the forum would be great, but it requires time and wise people :D

davidc@davidc-dell:/mnt/c/tmp/clean$ lualatex-dev test2
This is LuaHBTeX, Version 1.18.0 (TeX Live 2024)
 restricted system commands enabled.
(./test2.tex
LaTeX2e <2024-11-01> pre-release-3 (develop 2024-9-15 branch)
L3 programming layer <2024-09-10>

@MarcelKrüger that's taking a while...

@DavidCarlisle It shouldn't, this is not supposed to run anything that doesn't run in a normal latex run... Can you reproduce the delay? Is force rebuilding your font cache also extremely slow?

@MarcelKrüger sorry, fixed...

@PabloGonzálezL With the standard kernel definition of \item, \let is safe.

@egreg not with the tagging code. There we use \RenewDocumentCommand.

@UlrikeFischer I was explaining why \let is used in many answers.

@egreg I am sure every such answer was posted by someone who had checked the code and knew \let was safe:-)

While starting to write an answer on the main site I realized that with s.th. like (<kbd>A<sub>11</sub></kbd> a linebreak can occur between ( and <kbd>A<sub>11</sub></kbd>. How can such a linebreak be prevented? I tried inserting a word-joiner, ⁠, i.e., (&#8288;<kbd>A<sub>11</sub></kbd> to no avail.

does anybody know why my calculations are wrong?

I do think there must be a better way, but I also think I should be able to get this one to work. it almost does, finally, but not quite ....

is it possible to tell l3build or pdftex to quit after such-and-such time? (any method which works on linux would do. I don't care if it is part of tex.)