twitter.com/MLE_Online/status/1543329497113497601

usually the latex/LaTeX confusion is in the other direction :-)

LaTeX2e <2022-06-01> patch level 4
L3 programming layer <2022-06-16>

@JosephWright I suppose one option for @moewe would be to use the original \@uclclist of commands to uppercase by hand? vvvv

\documentclass{article}

\usepackage{graphicx}

\DeclareRobustCommand\fool{lower: \includegraphics[height=1ex]{example-image-a}}
\DeclareRobustCommand\foou{upper: \includegraphics[height=1em]{example-image-b}}
\makeatletter
\g@addto@macro\@uclclist{\fool\foou}
\makeatother
\begin{document}

\fool

\MakeUppercase{\fool}

\end{document}

@DavidCarlisle Could be

\documentclass[ngerman]{article}
\usepackage[T1]{fontenc}
\usepackage{babel}
\usepackage{csquotes}
\usepackage[style=authoryear]{biblatex}
\ExplSyntaxOn
\let\MakeUppercase\text_uppercase:n
\text_declare_expand_equivalent:Nn \bibstring \expandablebibstring
\def\expandablebibstring{HOLDER}
\ExplSyntaxOff

\addbibresource{biblatex-examples.bib}

\renewbibmacro{cite}{%
  \bibstring{bibliography}%
  \setunit{\addspace}%
  \def\expandablebibstring{UPPER}%
  \MakeUppercase{\bibstring{bibliography}}%
  \setunit{\addspace}%

texdoc.org is now at TL2022 :)

\documentclass[ngerman]{article}
\usepackage[T1]{fontenc}
\usepackage{babel}
\usepackage{csquotes}
\usepackage[style=authoryear]{biblatex}
\ExplSyntaxOn
\let\MakeUppercase\text_uppercase:n
\let\MakeLowercase\text_lowercase:n
\text_declare_expand_equivalent:Nn \bibstring \expandablebibstring
\def\expandablebibstring#1{HOLDER}
\ExplSyntaxOff

\addbibresource{biblatex-examples.bib}

\renewbibmacro{cite}{%
  \bibstring{bibliography}%
  \setunit{\addspace}%
  \def\expandablebibstring##1{UPPER}%
  \MakeUppercase{\bibstring{bibliography}}%

@moewe, @DavidCarlisle, @UlrikeFischer ^^^ Basic idea of how to set this up using the mechanisms built in to expl3

@JosephWright I need to add a patch to allow biblatex to load as I have PL4:-)

\documentclass[ngerman]{article}
\usepackage[T1]{fontenc}
\usepackage{babel}
\usepackage{csquotes}
\usepackage[style=authoryear]{biblatex}
\ExplSyntaxOn
\let\MakeUppercase\text_uppercase:n
\let\MakeLowercase\text_lowercase:n
\let\DemoSub\text_declare_expand_equivalent:Nn
\ExplSyntaxOff

\addbibresource{biblatex-examples.bib}

\renewbibmacro{cite}{%
  \bibstring{bibliography}%
  \setunit{\addspace}%
  \DemoSub\bibstring{UPPER}%
  \MakeUppercase{\bibstring{bibliography}}%
  \setunit{\addspace}%
  \DemoSub\bibstring{lower}%

@moewe, @DavidCarlisle, @UlrikeFischer ^^^ Slightly clearer version - we can use \text_declare_expand_equivalent:Nn to replace \bibstring by something else - I'd go for an expandable simplified version, as then it would 'just work'

@JosephWright do you need \let\MakeUppercase\text_uppercase:n here (losing the last remaining 2e wrapper?)

@DavidCarlisle That was just for a setup that worked with PL3 and would show the right effect - nothing to do with the fix

@JosephWright that's what I thought, just checking...

@DavidCarlisle, @UlrikeFischer (@moewe) I guess we might want a version of \text_declare_expand_equivalent:Nn that's 'case aware', so if provides an upper/lower/titlecase version

@DavidCarlisle well the problem is that they want to map \bibstring to different commands for upper- and lowercase and \@uclclist doesn't really support this as it assumes a symmetric pair.

@JosephWright so the l3 version of my \@uclclist example above, using gratuitously non-expadandable "text" would be this (which works fine in PL4)

\documentclass{article}

\usepackage{graphicx}

\DeclareRobustCommand\fool{lower: \includegraphics[height=1ex]{example-image-a}}
\DeclareRobustCommand\foou{upper: \includegraphics[height=1em]{example-image-b}}

\ExplSyntaxOn
\let\DemoSub\text_declare_expand_equivalent:Nn
\ExplSyntaxOff
\begin{document}

\fool

\DemoSub\fool\foou
\MakeUppercase{\fool}

\end{document}

@DavidCarlisle simply do \def\blx@err@patch#1{} % avoid error message

@UlrikeFischer having an example that doesn't use biblatex also works:-)

@JosephWright but as you say really wants a list of triples (or for Makeuppercase a list of pairs as in @uclclist) rather than flipping the general text expansion exception list betwn upper and lower

@JosephWright but presumably adding to the list slows down all usage so an inline version where you \def\foo{\loweruppertitle{foo}{FOO}{Foo}} and \text-.. pick the right one may be preferable for some usages

@DavidCarlisle I think that is useful too (perhaps even with four argument for the no-casing-at-all case). It is like \texorpdfstring and \pdfstringdefDisableCommands. One doesn't want to have all sort replacements for all odd commands in some fix list.

Anyone here which understands the tex c code? The miktex seems to have to do with a buffer overflow but I can't access which engines this can potentially affect and when: github.com/MiKTeX/miktex/issues/1121#issuecomment-1173039008

@UlrikeFischer well I can more or less read the fragment but not seeing the context yet. buf is of size 20 and handled differently for xetex and ptex (utf-16) but I don't see why (or where) it is using the file contents there, that's never going to fit in 20 bytes. I'd just forward to texlive list

@DavidCarlisle ooh a plan

@DavidCarlisle did the comment disappear again?

@UlrikeFischer ?

@DavidCarlisle the miktex comment about the buffer overflow. I just wanted to sent to the texlive list and it seems to be gone. You just made more or less the same changes I made too including the command name ;-)

@DavidCarlisle but I think you should change the order here

      \let\HDorg@M^^M%
      \catcode`\^^M=\active

@DavidCarlisle @UlrikeFischer Yes, I can sketch something up - the key for @moewe is that we will fix this I think at the expl3 end in a day or so (I'm busy most of today, so an extension may need to wait until tomorrow)

@DavidCarlisle I like that, but I wonder how @moewe hooks in - still seems to need a 'use a special definition in case changing' setup hook

@JosephWright is a hook needed couldn't biblatex define its commands to be \def\whatever{\CaseSwitch{normal}{lower}{upper}{title}} then like \NoCaseChange, \CaseSwitch has a global definition of \@firstoffour but within the text caser you make it pick the right branch

@UlrikeFischer I did it that way on purpose, to make it clear that the setting that matters is the one outside the definition

@DavidCarlisle but if ^^M isn't already active you are not storing/restoring the definition it has in the active case aren't you?

@UlrikeFischer oh I see what you mean about the miktex, I just refreshed the page and it went. maybe the comment was wrong, which was why I couldn't follow it:-)

@UlrikeFischer the current catcode setting isn't used, \let\HDorg@M^^M% saves the definition of an active ^^M, the following \catcode\^^M=\active` is for reading the input file, it has no effect on tokens within the definition

@UlrikeFischer that ^^M is known to be an active token, as it is set on line 744

@DavidCarlisle ooh

@PauloCereda breakfast^^M

@DavidCarlisle oh no^^M

@DavidCarlisle ah right.

@UlrikeFischer so I push to ctan?, they will be missing us, it's almost mid day and they have not had a ctan upload from us yet today

@DavidCarlisle the question is if one want \CaseSwitch in lots of command even if the document never use \MakeUppercase. So perhaps a additional hook, to allow also \let\foo\foocasing before the casing command if wanted and then one could do \foocasing= \CaseSwitch{normal}{lower}{upper}{title}.

@DavidCarlisle yes please, I will be away for the afternoon. Or perhaps one should try first with a few documentations ;-)

@DavidCarlisle L3 is the new easybook?

@DavidCarlisle oh, more $$...

@UlrikeFischer why is math being set in .\PU/pdf/m/n/10 1 ??

@mickep $$ least of the problems there:-)

@DavidCarlisle Oh I see. I like the argument by authority, though. :)

why when looking at the source of a package you have never heard of do you always find comments like % add 14apr18 per dcarlisle, tug.org/pipermail/tex-live/2018-April/041437.html

@DavidCarlisle We could use $$$

@DavidCarlisle Sure but we need to swap out what \bibstring does without I guess imposing that on biblatex. Or are you thinking \def\bibstring{\SomeCaseFunction}

@UlrikeFischer So something like a two-part setup: we have \CaseSwitch and also \text_declare_case_replacement:Nn, then we use the latter to allow people to swap out commands to case-switching ones (where we assume these are protected functions so don't just 'produce text')

@moewe I think we will have a solution today :)

@DavidCarlisle Which one?

@DavidCarlisle Isn't it nice to see your ideas/code being added to the packages?

@JosephWright Probably this question

@DavidCarlisle ??

@JosephWright mathtext but as @mickep says. that is actually a brilliant mwe, every line is needed:-)

@JosephWright sounds good

@DavidCarlisle if you move mathtext just before hyperref it stops also. That's a real puzzle.

@Rmano or add what it says in the warning and add \usepackage[T2A]{fontenc} it looks like something is adding a token after the closing $$ so defeating the implicit ignorespaces, so the end of line makes a space, but haven't found it yet

@PauloCereda better to use €

@DavidCarlisle ooh will \usepackage[british]{babel} catcode that to £? :)

@UlrikeFischer the question asks about a space after $$ but the $$123$$ text is set in PU encodig

@DavidCarlisle I'm, can't test now, but has it to do with the safe/restore spaceskip code of hyperref?

@UlrikeFischer I need to eat duck lunch, will come back to this later

@DavidCarlisle oh no

The hunt for space.

@DavidCarlisle I've logged in the issue

@UlrikeFischer @JosephWright on the space after math, ... \group_insert_after:N \mv@normal@reset I wonder who we can blame....

@DavidCarlisle ???

@DavidCarlisle Oh

@JosephWright you get a space after $$ because the new "undo mathalphabet assignments if you are running out of slots" code aftergroups a token that loses the ignore space after $$ so using \EuScript in an earlier expression causes a space to appear later in the document, trying to make a MWE

Can someone point me to the code from whence this warning comes from?

> LaTeX Warning: There were undefined references.

@FaheemMitha source2e, \G@refundefinedtrue

@MarcelKrüger OK. Can I find that in the source2e document?

I.e. texdoc source2e?

It looks like it might come from sources.debian.org/src/texlive-base/2022.20220605-1/texmf-dist/…

What is latex.ltx?

Judging by the hits I'm getting, it's possible to customize this warning. But I can't find any pointers on how to do so. Can anyone provide anything?

@FaheemMitha Sure, just look for ` \G@refundefinedtrue. latex.ltx` is the source of the LaTeX kernel, all of it is in source2e.

@MarcelKrüger OK.

@FaheemMitha it's the source of latex

@DavidCarlisle The entire thing in one file?

@FaheemMitha yes it is generated from 50+ documented source files, look at the comments at the top

@FaheemMitha you can only customise it in the sense latex is open source and you can change anything. It is not designed as a configuration point

@DavidCarlisle OK.

@DavidCarlisle OK. Searching for \G@refundefinedtrue brings up a lot of hits. I don't understand the code, but I thought maybe it was for customising that message.

@FaheemMitha it's just a tex primitive \if, it is not a macro with a definition. it is \iftrue if there has been an undefined \ref and \iffalse otherwise.

@DavidCarlisle OK.

@DavidCarlisle well it is the fault of this mathtext.sty, and it is old:

\documentclass{article}

\usepackage{mathtext}

\DeclareFontEncoding{PU}{}{}
\DeclareFontFamily{PU}{pdf}{}%
\DeclareFontShape{PU}{pdf}{m}{n}{ <-> cmr10 }{}%
\DeclareFontSubstitution{PU}{pdf}{m}{n}%

\begin{document}
abc
\showoutput
$a$

\end{document}

@UlrikeFischer why would it do that?

@DavidCarlisle no idea. I thought you know as you offered a solution in 2018 when it broke because of the utf8 move. ;-)

@DavidCarlisle do you know why \IeC is called like that? (question on the german list)

@FaheemMitha I answered your question, but received no feedback. Did it solve your problem?

@AlanMunn Oh. Yes, I think it did. Or will. I was trying to get to the point where I could confirm that it did, but I haven't yet.

@AlanMunn But I should at least upvote. I'll do that now.

@UlrikeFischer IeC is called like what? (just discussing with frank if we need PL5 for that space

@UlrikeFischer oh you mean why that name? InputEncodingCommand

@DavidCarlisle Good thing with many users: Also the funniest bugs are found.

@mickep I an ideal world I'd use a math typesetting engine only available to one or two people. could you imagine such a world?

@DavidCarlisle I don't get your point.

@DavidCarlisle Why would that be an ideal world?

@FaheemMitha users generate bug reports, feature requests,...

@DavidCarlisle Is that bad?