TeX, LaTeX and Friends

cfr

05:41

@UlrikeFischer but l3build usually only compiles once if that happens and returns an error level. it doesn't build a ctan archive.

@UlrikeFischer but I admit this looks odd ...

I really meant it doesn't load without error. it isn't a fatal error, but still I don't understand why turns out so differently.

cfr

06:07

@UlrikeFischer thanks very much for this.

Skillmon

09:11

@DavidCarlisle I'm shocked that you don't suggest the superior expkv-opt (tex.stackexchange.com/questions/736966/…)

David Carlisle

09:51

@Skillmon best to stick to standard interfaces, would you trust a de-furred rabbit to maintain your package interface?

Paulo Cereda

@DavidCarlisle quack

David Carlisle

@PauloCereda b

Paulo Cereda

@DavidCarlisle o

Skillmon

@DavidCarlisle expkv-opt is more flexible, imho. The only missing 1:1 feature is the .usage:n (which can be handled by different sets or key-undefining, or bulk-key-redefinitions). (but for professional projects I use ltkeys as well)

David Carlisle

10:07

@Skillmon yes but I think the mesage for "new user" was "don't use the non standard option handler, use the new standard one" which is a better sounding message than "don't use that non standard option handler, use a diffent non standard option handler" expert users can make an informed choice, but for a new user...

Skillmon

10:20

@DavidCarlisle ok, agreed

Joseph Wright

For people interested in expl3 consistency: github.com/latex3/latex3/issues/1682 thoughts welcome

egreg

10:56

@JosephWright So you're suggesting, say, \tl_map:nN and \tl_map:nn instead of \tl_map_function:nN and \tl_map_inline:nn? But how would \tl_map_tokens:nn fit?

Skillmon

@JosephWright whoopsie, commented with the wrong account... :) I removed my comment on GH, this is the content of it:

IMHO, deprecating map_function for map_tokens is fine.

The map_inlines, otoh., are handy and reduce hash-table flooding. If you deprecate these you always need to define an auxiliary for every loop, and if you re-use an internal need to keep track of scoping yourself.

Maybe reduce to <type>_map:[Nn]n for the tokens version plus <type>_map_inline:[Nn]n for the inline version.

David Carlisle

11:14

@Skillmon I just said something smilar (I think:-) in gh

Skillmon

11:26

@DavidCarlisle except for the "deprecate _function" part

David Carlisle

@Skillmon That was the "rationalise names" (more or less). You are free to repost under your preferred account (probably you should so people not here see it)

Joseph Wright

12:03

@egreg I'm suggesting \tl_map:nn as equal to \tl_map_tokens:nn and moving to that generally

mickep

12:13

@DavidCarlisle $$\begin{Large}Some beautiful math we see today\end{Large}$$

David Carlisle

@mickep I thought so

mickep

@DavidCarlisle The diff question was better, or would have been if it would have been about why different(yay!) results are obtained with different(yay again!) fonts, and not what is preferred to use.

David Carlisle

@mickep you mean the question "why if I use explicitly unsupported syntax that gives consistent but hard to predict and never desired behavour, do I get hard to predict and undesired behaviour?"

mickep

@DavidCarlisle Oh, what is unsupported? \mathit? \textnormal? I meant this one.

Skillmon

@DavidCarlisle I'd change the formatting of your $$ answer. This way the first thing someone sees as part of an answer is the evil wrong bad syntax.

David Carlisle

12:26

@mickep are sorry I thought you refering to the same \begin{large} question asking about diff (erences0 in behaviour

mickep

@DavidCarlisle Ahh, no, that question with the beautiful math is just... weird.

David Carlisle

12:39

@mickep speaking of beauty, you should be used to reading long words, so should like my last answer.

mickep

@DavidCarlisle That is the second nicest LaTeX document I've seen today, indeed.

Ulrike Fischer

13:08

@DavidCarlisle verrrykonng?

mickep

@UlrikeFischer He could just have given a simple German sample text.

Ulrike Fischer

@mickep or have used kingkongking ;-)

mickep

@UlrikeFischer Oh, that would also have worked.

egreg

@JosephWright We really need a dictionary “deprecated → preferred”

Joseph Wright

@egreg We can mark that way ...

Joseph Wright

13:35

@cfr This just came up at work: Llwybr datblygiad proffesiynol gyda hyfforddiant yn y gwaith :)

Non-apprentice apprentices ...

koppor

Today, I "quickly" wanted to compare tooling for a self-contained .tex file (refine my 2022 answer at https://tex.stackexchange.com/a/629364/9075 - to properly add DEPP).

But it turned out that I am stopped by something else. "minted" is not working any more (https://github.com/gpoore/minted/issues/435) and "cfr-lm" is not working in MiKTeX (https://github.com/MiKTeX/miktex-packaging/issues/456).

David Carlisle

@UlrikeFischer yes I noticed after I'd made the image and let it go it's not as ifff ttthhhee rrresst off the sppppellling was rigght

Joseph Wright

@koppor Oops

Jasper Habicht

13:54

Typing code examples on a mobile phone ist really horrible

3

Joseph Wright

@JasperHabicht :)

Jasper Habicht

@JosephWright I hope the code I just added to my comment on GitHub even compiles …

For whatever reaon I can’t even establish a hotspot from my phone … i really hope my provider can fix their issue quickly!

Skillmon

@JasperHabicht _map_function:[Nn]N can be swapped to _map_tokens:[Nn]n without any other code change.

cfr

@JosephWright how did that come up?

Joseph Wright

@cfr I'm in an online teaching meeting - we have pharmacy technicians doing apprenticeships - the ones from England are 'apprentices', the ones from Wales are not ... but do the same course ... all about money apparently

@Skillmon Quite, that's the bit that really stands out

Skillmon

14:06

(the only thing that's happening is that with n-type you can't use a simple \exp_after:wN in the implementation to expand something, so n-type either needs argument juggling (not so good performance wise) or some hack like \expanded{\unexpanded{#1}\expandafter}, so the map tokens variants weren't implemented historically and only creep into the language relatively recently)

@JosephWright just because of his comment on GH

> As long as a similar functionality is provided and a hint in the manual as to how to switch from function to whatever, this should not be a problem for package authors who currently make use of the function mappings.

cfr

@JosephWright ah. I thought it might be because the alternative rolls so easily off the tongue.

Joseph Wright

@cfr :)

Jasper Habicht

@Skillmon Cool, then I‘m totally fine :) I thought that _token does something else (without exactly knowing what)

I just thought about the guys who currently use function and don’t know much about the other mapping functions … such as me …

Sorry for the noise if this is not contributing much to the discussion

cfr

@koppor as far as I can tell none of my font packages work because MikTeX doesn't use the ins. but cfr-lm 1.8 is broken anyway. (1.8a works. 1.7 works.) but I still don't understand how l3build ctan produced a zip ....

yo'

@JasperHabicht I am regularly using Overleaf on a cellphone, and that's yet another level of challenge!

Jasper Habicht

14:20

@Skillmon I should have read the manuals in more detail … it even says that _tokens is equivalent to _function if only one token is given in the argument (which can be a function, I assume)

Skillmon

@JasperHabicht the difference is that tokens accepts more than a single function, so you can for instance use \clist_map_tokens:Nn \l_tmpa_clist { \my_function:nn {abc} }

Jasper Habicht

@Skillmon And then the second argument to \my_function would be the item from the clist, right?

Then it is indeed a good idea to unify! It is actually indeed the very same thing

Joseph Wright

@JasperHabicht Yes

Skillmon

@JasperHabicht except that \clist_map_function:NN \l_tmpa_clist { \my_function:nn {abc} } fails it is the same thing, yes :) (the implementation of \keyval_parse:NNn is for that reason a simple \cs_new_eq:NN \keyval_parse:NNn \keyval_parse:nnn)

Jasper Habicht

@Skillmon I see! Then _map_function:NN could become _map:NN for example …

Or just stick to :Nn and pass the function as token in braces

Skillmon

14:36

@JasperHabicht well, the point is that we don't need the _map_function:NN anymore and can simply no longer document them. So we only provide map:Nn as the tokens variant, and map_inline:Nn as the inline variant, and all other versions silently hide in the back of the documentation.

Performance might degrade a tiny bit, but that shouldn't be a big hit and is most likely negligible as soon as someone does typesetting.

Jasper Habicht

@Skillmon Yes. I understand the idea

Skillmon

@JasperHabicht what I mean is that "_map_function:NN could become _map:NN for example" won't apply. They'll just cease to exist (in the documentation, the implementation will stay without any renaming).

Ulrike Fischer

14:56

6

Happy Birthday Grommit!

Jasper Habicht

16:18

@Skillmon yes, I meant that maybe one could establish an :NN variant of the new/old/renamed function which would exactly be the old _map_function:NN command. But of course this is not absolutely necessary … :Nn would also always work.

David Carlisle

This is pdfTeX, Version 3.141592653-2.6-1.40.27 (TeX Live 2025) (preloaded format=pdftex)

cygwin texlive 2025 now has new executables:-)

Joseph Wright

16:35

@DavidCarlisle Yay!

mickep

@DavidCarlisle Oh, very good. With latest additions/fixes (luatex)?

David Carlisle

@mickep not looked yet (it just arrived) but I just updated the linux one and that is still showing log indent diffs:

***************
*** 27,33 ****
  .\OT1/cmr/m/n/10 į
  ! OK.
  <to be read again>
! }
  l. ...{\showoutput\showbox0}
  > \box...=
  \hbox(6.63332+1.94397)x17.21802, direction TLT
--- 27,33 ----
  .\OT1/cmr/m/n/10 į
  ! OK.
  <to be read again>
!                    }
  l. ...{\showoutput\showbox0}
  > \box...=
  \hbox(6.63332+1.94397)x17.21802, direction TLT
***************

mickep

@DavidCarlisle Oh, but did anyone report that problem to the developers?

David Carlisle

@mickep I thought it was this?

yesterday, by mickep

@DavidCarlisle [luatex] A bug in show_context second line indentation

so I didn't try to extract a small plain tex test from the latex test suite

mickep

16:57

@DavidCarlisle I guess so too, but I did not report it. I thought you did.

David Carlisle

@mickep no sorry I thought you meant Hans had confirmed it's a bug. I'll make a test file later

@mickep it's confusing if this chat is used for discussing tex, we should stick to cricket and rugby

mickep

17:54

@DavidCarlisle good!

@DavidCarlisle indeed, maybe that is why tonight's context meeting is not taking place here.

Skillmon

18:49

@DavidCarlisle did you manage to get NeoVim running in cygwin?

David Carlisle

19:05

@Skillmon as well as needed: bash: neovim: command not found

Skillmon

19:19

@DavidCarlisle that's to be expected, the executable is named nvim

@DavidCarlisle more serious question: How well can Windows native executables be mixed into Cygwin? Would they pick up the correct environment (so Cygwin, if called from within Cygwin)?

David Carlisle

@Skillmon I hav ethat covered as well bash: nvim: command not found ; bash: vim: command not found

@Skillmon it's kind of a black art. there was never a cygwin java (for some reason) so I regularly use the saxon xslt systemit mostly works fine but you need to be careful passing in files as /foo/bar to cygwin is c:/cygwin64/foo/bar or even c:\cygwin64\foo\bar to native windows applications, so it takes some juggling of CLASSPATH etc to get things lined up.

@Skillmon I'm trying to wean myself off cygwin and switch to wsl2 (I have both installed) wsl2 is almost certainly the future rather than cygwin but....

@mickep report sent (to texlive and dev-luatex lists)

mickep

19:35

@DavidCarlisle Good!

David Carlisle

21:16

@mickep I should just have blamed @UlrikeFischer and not spammed the lists.

mickep

@DavidCarlisle Oh, well, no harm done, I guess.

Ulrike Fischer

@DavidCarlisle what did I do?

David Carlisle

@UlrikeFischer do you need to do something now to get blamed? whatever changed?

@UlrikeFischer If you need a plausable reason: you didn't tell me that the changed luatex logs were making them consistent with classic tex so we need to update the test suite, rather than blame Hans for breaking it:-)

Ulrike Fischer

@DavidCarlisle I hoped it was a bug report from me that triggered the change

mickep

@DavidCarlisle I don't think that change came from Hans.

David Carlisle

21:33

@mickep Hans has the honour of being blamed for any luatex change:-)

Jasper Habicht

21:49

Hooray, Cologne is back online

(it's been four hours now, but we have been offline for eight hours)

egreg

@JasperHabicht I'll drink some Altbier to celebrate!

Jasper Habicht

@egreg =3 I am originally from Krefeld, so you can't shock me with that

Skillmon

22:05

@DavidCarlisle well, let's just say I don't have a say in whether wsl2 is running on that pc or not.

@DavidCarlisle thanks :)

David Carlisle

`cygpath` is your friend various options to map between cygwin and windows paths: $ cygpath -m /tmp
C:/cygwin64/tmp

Skillmon

@DavidCarlisle I currently have a rather cumbersome setup with MSYS2/MinGW but have to switch laptops so was thinking about giving cygwin a try

koppor

22:19

Is it possible to create luatex-cache for all fonts on a system before any run?`

Changed files or newly in use/created:
/usr/local/texlive/2024/texmf-var/luatex-cache/generic/fonts/otl/lmroman10-bold.lua
/usr/local/texlive/2024/texmf-var/luatex-cache/generic/fonts/otl/lmroman10-regular.lua
/usr/local/texlive/2024/texmf-var/luatex-cache/generic/fonts/otl/qtgraphlite.lua
/usr/local/texlive/2024/texmf-var/luatex-cache/generic/names/luaotfload-lookup-cache.luc

context: https://github.com/dante-ev/docker-texlive/issues/51

Side story:

:/usr/local/texlive/2024/texmf-var/luatex-cache/generic# luaotfload-tool --find "test"
...4/texmf-dist/tex/luatex/luaotfload/luaotfload-arabic.lua:19: attempt to index a nil value (global 'node')

egreg

@JasperHabicht I have a friend in Krefeld, I've met him (he lives in St. Tönis, now) two years ago, before the TUG conference at Bonn. Exactly 50 years from my first visit and acquaintance with Altbier!

Jasper Habicht

@egreg Haha, it is so strange! Krefeld is really not a very large or famous city, but for some reason I often meet people who know someone from there ... I hope you liked it (although it has not very much to like ...)

Krefeld, I mean. The beer, I like actually ...

egreg

@JasperHabicht Quiet town with even something to visit: Burg Linn is interesting. Of course Düsseldorf has much more.

Jasper Habicht

@egreg Yes, Burg Linn is nice and the old town nearby! Düsseldorf is really too close, so almost no nightlife in Krefeld.

egreg

22:43

@JasperHabicht Much less in St. Tönis! 😉

Jasper Habicht

@egreg Oh, well yes ...

Transcript for

$TeX - LaTeX$ TeX, LaTeX and Friends