@Skillmon The constant updating seems to have disappeared. Maybe you could tell your fellow student to try delinking and relinking their Dropbox.

@AlanMunn I'll see him today, thanks for the update on your issues.

@DavidCarlisle, @UlrikeFischer, @egreg, @MarcelKrüger, @PhelypeOleinik (@PauloCereda) I see we are Agile

@JosephWright ;-)

@JosephWright from fragile to agile ...

@UlrikeFischer It's been obvious for a while: I think it might be my/Will's fault

@JosephWright ;-). I'm just breaking all hyperref tests with luatex ...

@UlrikeFischer Sounds fun

@UlrikeFischer I'm tracking down all of the files we need to build the format ... for ... secret reasons

Some years ago I read about a non-converging LaTeX document, i.e. one where a label and its ref are right at the page boundary, such that with every time you rerun, it jumps to the next of previous page. This was published in DTK I believe, but I can't find it in the archives. Can somebody help?

@UlrikeFischer ^^^

@FaheemMitha github.com/pgf-tikz/pgf/commit/…

@HenriMenke :)

@HenriMenke Thank you.

I was just told that I can edit pages on GH directly, and a PR will be created for me.

@JosephWright My student do this all the time and they are native speakers.

@FaheemMitha Yes, that is indeed possible. When you browse files, there should be an edit icon in the upper right corner.

@HenriMenke I'll keep that in mind. Thank you.

@FaheemMitha Yes

@FaheemMitha That's why GitHub is popular

@HenriMenke Yup, it's a tricky one

@JosephWright Plus all the pretty shiny colors.

@FaheemMitha Sure, but it's the feature set that drives use, really: it's been a big bonus to us (LaTeX team) compared with the previous set up we used

@DavidCarlisle ^^^ Gnats ...

@JosephWright CVS?

@FaheemMitha Subversion was not so bad ... Gnats however ...

@FaheemMitha ^^^

@HenriMenke Yes, I see it. I'll attempt to fix other typos directly, assuming I come across any.

Thank you again.

@FaheemMitha Linus Torvalds said that before he wrote git, the kernel was managed using diffs and patches via email, which he thinks was still a superior version control system to CVS.

@HenriMenke I think he might have been overstating things. Even a not-so-good VCS is better than none at all.

@JosephWright My favourite one so far was “conjoget” instead of “conjugate”.

Though I never used CVS. I had difficulty figuring it out. The first VCS I used was Subversion, in 2003 or so. Rather late in the day, really.

@JosephWright The LaTeX project switched from CVS to Subversion?

Did you use Savannah?

@FaheemMitha The TeX Live repo is still in Subversion, but that is because the repo is absolutely enormous and git actually can't handle it.

@HenriMenke Oh

@FaheemMitha Well, it has tons of binary files. All the fonts and PDF documentation for packages. And it's been going for at least 10 years, so there is a lot of stuff there.

@HenriMenke The PDFs are under version control? Don't they get built from source?

@HenriMenke I have one on this site

@FaheemMitha originally RCS then CVS then SVN now git

@JosephWright gnats is lovely

@FaheemMitha not in general, if authors submit a pdf to ctan it goes to texlive as-is

@FaheemMitha no, prior to github we used dante's server for a self-hosted repository

@DavidCarlisle Thanks David!

@JosephWright hm?

@PauloCereda handling bug reports in less than 10 years seems pretty agile to me.

@DavidCarlisle ooh

@JosephWright github.com/Cosmo/Clippy

@PauloCereda twice as many stars as arara

@DavidCarlisle Sorry I'm probably still asleep...

@DavidCarlisle :(

@DavidCarlisle That's unfortunate. Since the PDF is presumably generated from TeX source, ideally only the source should be under version control. And there should be build testing.

@FaheemMitha there are essentially no requirements on people uploading to ctan. A requirement that every package had to have a build system compatible with the texlive makefiles would be an insurmountable bar to most people who submit.

@FaheemMitha not to mention that it would mean that it takes weeks to build texlive from a clean checkout.

@FaheemMitha it takes some hours just to make all the pdfs that I submit, never mind all the thousands of other packages.

@DavidCarlisle Does that mean that the source of the PDF as well as the PDF itself is uploaded?

@FaheemMitha If you want it to go into TeX Live then yes: CTAN have no formal requirement

@JosephWright I see.

Does TeX Live commit PDFs too?

@FaheemMitha The TeX Live version control is only really there to allow Karl et al to stay sane: it's not so much for 'development'

@FaheemMitha Yes, otherwise they would not be available!

@JosephWright So TeX Live doesn't build them either?

@FaheemMitha For example, I build LaTeX2e as a ready-to-install zip: it's used as is to create the TeX Live package

@FaheemMitha No

Oh. In that case, I wonder how Norbert manages. Because Debian requires things to be built from source.

@FaheemMitha if Karl finds a pdf without source he removes it (or gets the author to add the sources) but basically relies on the authors to say everything needed is there, but whether it uses latex or context or xetex or requires three passes and xindy or ... to build is not recorded in general.

@DavidCarlisle I see.

@DavidCarlisle I broke lots of hyperref tests ;-) (I made in luatex unicode the default).

@UlrikeFischer ooh

@DavidCarlisle I wonder what to do with github.com/ho-tex/hyperref/issues/93. One can suppress the space is the first argument is empty, but this changes spacing also if the field is after some text in the middle of the line. Currently one get two spaces there:

\documentclass{article}
\usepackage{hyperref}
\usepackage{showframe}

\begin{document}
\begin{Form}
\noindent\TextField[name=a,width=2cm,charsize=8pt, height=.5cm]{}
sometext \TextField[name=a,width=2cm,charsize=8pt, height=.5cm]{}
\vspace{4ex}

\def\LayoutTextField#1#2{% label, field
  \def\temp{#1}%
  \ifx\temp\empty
   #2%
  \else
  #1 #2
  \fi
}

\noindent\TextField[name=a,width=2cm,charsize=8pt, height=.5cm]{}
sometext \TextField[name=a,width=2cm,charsize=8pt, height=.5cm]{}
\vspace{4ex}

\end{Form}

suppress space, I think. @UlrikeFischer

@DavidCarlisle and take the heat because of the changed documents? OK. How do one do the \ifx empty test correctly in latex2e speach?

@UlrikeFischer existing documents with a double space are probably in error even if the authors hadn't noticed? Testing empty as you have or \if\relax\detokenize{#1}\relax if you want to be expandable (but that probably doesn't matter here.

@DavidCarlisle yes, it is certainly more logical to suppress the space, but probably someone will complain anyway ;-) I'm redoing the luatex tests and then will add the change too.

@UlrikeFischer it's OK because if you do the commit, I can blame you.

@DavidCarlisle git commit -m"as wished by David Carlisle" ?

@DavidCarlisle twitter.com/mattdsteele/status/1173628386742345728

A had a request for mathtools to make \MoveEqLeft ignore the argument if used as \MoveEqLeft [3] (adding spaces). As far as I can see that is not possible, even with xparse. I was reading in the xparse manual, I don't think it is obvious from the xparse manual that !o does work with envionments while it does not with normal macros.

@daleif Can't be done: TeX ignores spaces after a control word

@JosephWright you update, I sync the branch. :)

@JosephWright I know, but it works for environments, but it is not obvious from the manual that it works for environments.

@daleif Hmm, OK: I guess we need to tighten up on the docs there

@daleif it would work with macros if not the first argument as it's just after the csname that you do not get space tokens

@JosephWright one strange thing that I do not fully understand. In mathtools at the very end of the sty, Morten added three lines that makes aligned not ignore spaces on the optional arg. But that applies to a macro. I'm guessing it has to do with the fact that again envs are at play

@daleif What's the code there?

@daleif amsmath is strange (or broken depending on your point of view) in that it changes \\ so that a newline stops the lookahead for the optional argument but a newline at the start after \begin{whatever} isn't treated specially and the optional argument can be on the next line

@DavidCarlisle Yeah, that's ... odd

@JosephWright it comes from that maybe spaces off macro, it ends up pointing to a macro in mhsetup that redefines @xargdef and then it stopped making sense to me.

The same construction are used on all the matrices in their starred variants

@daleif One for xmaths (@egreg)

@daleif Does that code do something to begin with?

@egreg which code? (on my way out the door, back later, all from tables, thus I haven't posted any code from mathtools). If I'm not mistaken the relevant code is at the bottom of mhsetup.sty)

@daleif The code at the bottom of mathtools.sty.

\def \MH_group_align_safe_begin: {\iffalse{\fi\ifnum0=`}\fi}
\def \MH_group_align_safe_end:   {\ifnum0=`{}\fi}

@daleif ^^^ ?

@daleif If so, it's 'Appendix D', nothing to do with the align environment really

@JosephWright I think the reference was to

\MaybeMHPrecedingSpacesOff
\renewcommand\aligned@a[1][c]{\start@aligned{#1}\m@ne}
\MHPrecedingSpacesOn

@egreg I wondered ...

@egreg Sort-of poor man's xparse: we should get Morten back properly ;)

theverge.com/2019/9/17/20870050/…

@DavidCarlisle ouch

@DavidCarlisle Oh .. dear

@JosephWright the news are really disturbing

@PauloCereda clearly we should switch to github actions from travis ci

@DavidCarlisle ooh

@egreg exactly

Interesting quote:

> [...] [However,] the approach of “just do what language X does” is intrinsically irresponsible; nearly every feature of every language is conditioned by other features of that language. Instead, the game is to learn from how other languages do things, assess the tradeoffs they’ve chosen (explicitly and implicitly), and ask what can be applied to the constraints of the language we have and user expectations within the community we have.

any miktex users around? could you check what you get for

@DavidCarlisle will look.

@UlrikeFischer the OP showed the log, and has a file in the working directory so it's not a miktex issue.

@DavidCarlisle yes, just saw it.

@UlrikeFischer I would guess some conference modified the class without bothering to change the name or version number or date or anything else that might give a clue:-)

@UlrikeFischer, @DavidCarlisle Getting close to having format-building as part of expl3 testing

@JosephWright isn't that a secret?

@UlrikeFischer That's not a secret ... it will speed up testing, that's all ;)

@UlrikeFischer Revealed the need for a new l3build feature

@JosephWright oh misunderstanding ;-)

@JosephWright ?

@UlrikeFischer Building a format needs some kind of hook just before running tests; the 'ad hoc' approach used just for LaTeX2e doesn't scale so well

@UlrikeFischer There are a few aspects of the LaTeX2e tests that I should tidy up to use newer l3build features

@JosephWright regarding l3build feature, could one have options --use-dev and --no-use-devor something like that? One can do it with build.lua and configuration files, but sometimes it would be nice to do it on-the-fly. (could be difficult to get the order right, depending on the build.lua so I'm not sure if it really can work).

@UlrikeFischer Perhaps we need a broader feature: some kind of list(s) to show 'standard branches' and 'dev branches'

@JosephWright that's why I cannot have nice things. :)

@JosephWright I have a secret plan :)

🇬🇧

Good evening. Is this number a good luck charm, a lucky number, or do I have to play superenalotto or lotto? :-)

@Sebastiano some of the grumpy people here will complain the comma is in the wrong place, but congrats you have lots of made up internet points.

@StrongBad who would do such a thing?

@DavidCarlisle what is your take on the asymmetry of the comma as a separator?

@StrongBad What comma?

@Sebastiano Palindromes are always welcome!

@DavidCarlisle the invisible one that is throwing off the kerning

@StrongBad my view is that U+2063 shouldn't be considered as breaking a palindrome

@MarcelKrüger I'm just doing the last steps for a luaotfload update (v3.00) based on the current dev. I changed the build lua a bit, most importantly I thought we should separate the package date from the fontloader date, so it has now both.

@DavidCarlisle Look how Christian commented on my miktex issue github.com/MiKTeX/miktex/issues/369.

@egreg It'll bring me luck though. Anyway, always thank you.

@StrongBad @DavidCarlisle @egreg The important thing is that you don't quarrel about my palindrome. I just wanted to joke with this number or as we say in Italy, "ridere sarcasticamente sulla mia sfortuna". Surely egreg can translate it better. I'm interested in a bit of luck :-)

@UlrikeFischer Will you be able to review and test this anytime soon? github.com/pgf-tikz/pgf/issues/743

@UlrikeFischer grrr, but commandline processing is explicitly system dependent so he can actually make it do whatever, but there is really no reason to make systems differ here and since it works everywhere else.....

@UlrikeFischer if you take that view you wouldn't be able to use any tex commands on the commandline which would be severely limiting

@DavidCarlisle well I tried a few commands and they work. So it looks like some heuristic. And it is not consequent: 5%.tex is a valid file name too and it fails.