TeX, LaTeX and Friends

mickep

12:27 AM

@DavidCarlisle Well, now TL24 is released, at least. tug.org/pipermail/tex-live/2024-March/050210.html

David Carlisle

12:43 AM

@mickep installled here:-)

mickep

@DavidCarlisle Oh, very quick.

David Carlisle

@mickep well I had the pretest so I just had to switch repository to a normal mirror (I used a German one as the UK ones haven't picked up yet)

Skillmon

7:57 AM

@DavidCarlisle ooh, German efficiency.

Joseph Wright

l3sys-query sent to CTAN!

Paulo Cereda

8:33 AM

David Carlisle

@JosephWright all tests failing as gh is picking up tl2024 :)

Paulo Cereda

@JosephWright oh no

@Skillmon ooh

David Carlisle

@PauloCereda l3sys-query ls breakfast

Paulo Cereda

@DavidCarlisle cannot list 'breakfast': Permission denied

I mean

@DavidCarlisle oh no

Joseph Wright

@DavidCarlisle Yes, or rather as I've zapped the caches we have to deal with things that have change dep-wise: xparse in particular

David Carlisle

8:37 AM

@JosephWright you only sent it once? slacker.

Paulo Cereda

@DavidCarlisle ah easybook, we are back on track :)

@samcarter ^^

Ulrike Fischer

8:59 AM

@DavidCarlisle is gh switched automatically? I thought we have to actively change the texlive version?

Joseph Wright

@UlrikeFischer It's manual

samcarter

@PauloCereda what would ctan do without easybook? :P

Joseph Wright

@UlrikeFischer What happens when TL ticks over is the update process fails, so the runs fail, then we have to zap the caches and re-run

David Carlisle

@UlrikeFischer it does a tlmgr update and now errors because that reports the upstream repo is now tl2024. if you just delete the cache (or @JosephWright wakes up earlier and deletes it) then re-running the workflow gets back in sync

Joseph Wright

@samcarter Have you looked at the Git repo? There's no history or any code - v odd

samcarter

9:03 AM

@JosephWright wow, that's strange

Ulrike Fischer

@JosephWright and how does one switch? It would be nice to have only one version again, currently I have to pay attention to update tests with tl2023 as they are small changes in engines.

Joseph Wright

@UlrikeFischer It's a question of which mirror is being used: didn't Marcel adjust things so that's selectable?

samcarter

Oh, chocolate! omny.fm/shows/daniel-and-jorge-explain-the-universe/…

David Carlisle

9:19 AM

@UlrikeFischer you can't really stay on tl2023 unless you rely on the cache as all the ctan mirrors will be at 2024

@JosephWright I suppose an expl3 update then if the PR is merged at tools we could do a latex-dev/tools update (unless there is wider latex-dev update planned soon?)

Joseph Wright

@DavidCarlisle Yes, that's the plan - but currently having an issue with one failing test

Rmano

Ups. I tried to do the last update to TL2023 and it failed horribly:

[ 1/34, ??:??/??:??] update: asymptote [2681k] (70496 -> 70583) ... TLPDB::_install_data: downloading did not succeed (download_file failed) for mirror.ox.ac.uk/sites/ctan.org/systems/texlive/tlnet/archive/…
tlmgr: Installation of new version of asymptote failed, trying to unwind.
tlmgr: Restoring of old package did NOT succeed.
tlmgr: Error message from unpack: don't know how to unpack
tlmgr: Most likely repair: run tlmgr install asymptote and hope.
done

...and now I have a broken tl2023. Is it a mirror problem?

David Carlisle

@Rmano oxford is not synced yet the archive directory hangs and the TEXLIVE_2024 file is not at the top. Try ftp.fau.de/ctan/systems/texlive/tlnet

@Rmano or anorien.csc.warwick.ac.uk/mirrors/CTAN/systems/texlive/tlnet if you want to stay in the UK

AlexG

9:36 AM

Is there a way to query the current repo URL with tlmgr in TL23? I want to change the TL24-pretest one into the one I was using with TL23. I'd like to avoid re-installing TL24. This should be possible because I did it last year, but I don't remember.

Ulrike Fischer

@AlexG tlmgr option show

AlexG

@UlrikeFischer Cool! Thank you!

David Carlisle

@AlexG yes you can just change the repository with the repository option although I can never remember the option name so just change the line in /usr/local/texlive/2024/tlpkg/texlive.tlpdb that has the URL (you probably are not supposed to do that:-)

Rmano

@DavidCarlisle thanks. I'll recover from a backup, the failed update above deleted my install-tl and kpsewich... :-O

Or maybe not, and it was an unrelated problem... buf

AlexG

@DavidCarlisle That's easy with tlmgr, e. g. tlmgr option repository https://mirror.ctan.org/systems/texlive/tlnet. ("Examples" section at the beginning of the man page. :) )

David Carlisle

9:41 AM

@AlexG man page, what's that? but if you'd have opened the file I suggested for your tl2023 one you'd have seem your old url without having to ask or read the man page to find option show :-)

AlexG

@DavidCarlisle I know, documentation is often confusing. First I was trying tlmgr info which did not tell me the repo.

Ulrike Fischer

@DavidCarlisle but then I couldn't have shown that I just found that out too ...

@DavidCarlisle well it seems as if something happened without me doing anything. The tagpdf tests used tl2024 on gh, so I can now update them to this version.

Rmano

So, basically my fault. I used the --self in tlmgr and it seems that it was the culpable, now without it just says it's at 2024 and refused the update. Is there a way to update my 2023 to the very last version? (I have been lazy lately...)

David Carlisle

@UlrikeFischer @JosephWright I was just going to ask why it's not finding TeX Gyre, I see you changed the test but why doesn't the font name work?

Joseph Wright

@DavidCarlisle My guess is we need to re-run luaotfload-tool, but I'm not sure how to do that

Ulrike Fischer

9:52 AM

@DavidCarlisle sorry? which test?

David Carlisle

@UlrikeFischer github-0282 but any checkin (eg my tools PR) to the latex2e repo was failing as that one fails as it couldn't find TeX Gyre Termes

Ulrike Fischer

@DavidCarlisle ah latex2e. Perhaps it has to do with the new security settings? You should ask @MarcelKrüger

Joseph Wright

@UlrikeFischer Didn't think of that: could be I suppose, though works fine locally

Ulrike Fischer

@JosephWright but actually if it can use it by file name the cache must be there. So it is not a problem to generate it. But looking into the log: luaotfload looks quite outdated.

Joseph Wright

@UlrikeFischer Ah, could be that

Ulrike Fischer

10:03 AM

@JosephWright and lualibs too.

Rmano

@DavidCarlisle Wow, the fau.de one is really fast from here. Installing going on!

Paulo Cereda

@Rmano FAU is the best!

@samcarter ^^

samcarter

@PauloCereda Florida Atlantic University? :P

Paulo Cereda

@samcarter oh no

ooh alligators

Joseph Wright

@UlrikeFischer I'll update them, check for passing, then revert the change in the test and see what happens

Joseph Wright

10:30 AM

@UlrikeFischer, @DavidCarlisle I think I'll clean up the texmf dir generally - I think some stuff is no longer needed (as it's only for when tree searching is not active)

Skillmon

@DavidCarlisle seems easy to process

Joseph Wright

@DavidCarlisle Building latex3 now

A full set - CTAN will love me!

Skillmon

@JosephWright there are some activities, but just force-pushes of tags to the README...

Joseph Wright

@Skillmon Yes, quite - weird

Ulrike Fischer

@JosephWright what do you want to remove? While it is work to update the tree it also separates the latex2e tests from all the local development files.

Skillmon

10:37 AM

@JosephWright I wonder why no code is hosted there... I guess using it for issues and discussion ought to be enough for the author?! Strange nonetheless.

Joseph Wright

@UlrikeFischer I wonder about e.g. KOMA - if its in texmf then it's always available even if tree searching is disabled - just want to check we've not got stray files, really

@UlrikeFischer I see we have tagpdf there but also installed using tlmgr - presumably we want only one

Paulo Cereda

➜ tlmgr option repository
Default package repository (repository): ftp.fau.de/ctan/systems/texlive/tlnet

@Rmano @samcarter ^^ :)

David Carlisle

@JosephWright some latex-lab tests might need the non search tree? but in general rather than having copies in the local texmf that then get stale couldn't we symlink individual directories to the full tree so they are always in sync?

samcarter

@PauloCereda :)

Paulo Cereda

ooh TL 2024 here

Ulrike Fischer

10:52 AM

@JosephWright well I think that having there a clear, known version of tagpdf and the pdfmanagement is useful. If I locally update tests in latex2e I don't want to use the local versions in which I'm just doing some experiments. If we remove them from the tree can one then setup l3build so that it ignores my local install tree?

Joseph Wright

@DavidCarlisle Would that work on Windows? Only Administrator can make links

@UlrikeFischer If tree searching is disabled, that's what happens - what I want to know is why we have both a local copy and an install using tlmgr - it's that which is odd

David Carlisle

@JosephWright I meant at gh actions, but locally as well, hmm copies it is then

Joseph Wright

@UlrikeFischer No issue having a local copy - after all it's what we do for luaotfload for exactly that reason

Ulrike Fischer

@JosephWright well duplications happen ;-).

@JosephWright but we should keep them more in sync. expl3, tagpdf and pdfmanagement looks ok, but e.g. graphics-def seems to be outdated too.

Joseph Wright

@UlrikeFischer Yes, all true - I will do a cleanup

@UlrikeFischer You'd prefer we keep the texmf version of tagpdf and pdfmanagement, I take it - in which case, they should go from the tl_packages list (and I'll rebuild the cache)

Joseph Wright

11:23 AM

@UlrikeFischer Quite a few outdated, and each needs to be checked :)

Ulrike Fischer

@JosephWright the question is why they some of them are there at all. We should perhaps add a note ...

Joseph Wright

@UlrikeFischer Perhaps - I am working on the ones I have a reasonable idea about first :)

David Carlisle

11:57 AM

@JosephWright @UlrikeFischer coming back to symlinks (or copies) I still wonder if we should lose the texmf tree from git and instead have the build script just copy the files from the full tree to make a local tree (always, or if they are newer) so that the local texmf keeps in sync.

Actually it wouldn't have to copy it could just use a custom TEXINPUTS that picks out certain directories

Joseph Wright

@DavidCarlisle The original idea of the local tree was partly about when tree searching is disabled but also so we have predictable versions installed - for example, I updated l3kernel there for 2e before a release so we know that there won't be issues when it hits TeX Live

@DavidCarlisle True, we could add more refinement here: one issue is that really this stuff is 99.9% for the kernel - other than LaTeX itself, disabling tree searching and picking out packages one at a time is ... unusual

@DavidCarlisle How would you set that up? The current arrangement has the advantage it's quite easy to do at the Lua end - you imagining a table of paths?

David Carlisle

@JosephWright yes I know but it means stuff gets old and we end up testing against a custom setup unrelated to the setup that users use, eg graphics tests really ouught to be using the same graphics-def files as actually get used,

@JosephWright moan at you and you set it up was my current plan (I haven't looked in detail at all:-)

Joseph Wright

@DavidCarlisle Something like texmfinput = {"foo/bar", "bong/baz"} that we then concatenate into the input to TEXINPUTS?

Needs to check the install directory to set that up - I guess easy enough

@DavidCarlisle If you remember, the current approach is an evolution from the previous one where we had to copy every file from support :)

David Carlisle

@JosephWright yes more or less something like that, but maybe it wouldn't work some things are simple to restrict just using TEXINPUTS but anything with fonts use things from all over and you could end up with a broken setup if you restrict some of the paths and not others. But I don't know it just occurred to me now, wasn't a well formed plan...

@JosephWright yes but support only had half a dozen files if I remember rightly but things are more complicated these days.

Joseph Wright

@DavidCarlisle Indeed :)

Ulrike Fischer

12:08 PM

@DavidCarlisle yes, but it also allows to test new stuff (on github) without hurting the users first. I mean I can simply create a branch and move in a new tagpdf version and sent to github.

Joseph Wright

@UlrikeFischer Yes, I think we need the current mechanism at least for a few development files - @DavidCarlisle is thinking stuff where we just happen to need it with tree searching disabled (like unicode-data, say)

David Carlisle

@JosephWright yes

Joseph Wright

@UlrikeFischer, @DavidCarlisle I was right about KOMA: we don't need the local copy (I think we added it but then it was also included in the general install)

David Carlisle

12:25 PM

@JosephWright is it expected my PR checks still failing looking for TeX Gyre?

Check failed with log file
  - ../build/test-config-TU/github-0282.log

Joseph Wright

@DavidCarlisle Did you cherry-pick the luaotfload update?

David Carlisle

@JosephWright er no I just touched tools, I was thinking the rest would come from the changes you just made but I should rebase (or cherry pick) against ~~main~~ develop first as I'm running on the branch....

Joseph Wright

@DavidCarlisle Indeed

David Carlisle

@JosephWright so develop should be good?

Joseph Wright

@DavidCarlisle Yes

David Carlisle

12:34 PM

I wish github would show github.com/latex3/latex2e/compare/develop..l3sys-query not github.com/latex3/latex2e/compare/develop...l3sys-query by default (seems lots of peopel have wished that over the years)

Joseph Wright

@DavidCarlisle Yeah, does seem odd: I don't think you can say either is 'preferred' at a fundamental level

David Carlisle

@JosephWright it's a pain as it makes it look like there are lots of changes to review, seems the reason they don't do that is that review comments on commit diffs would get orphaned if a rebase made the diff vanish

Joseph Wright

@DavidCarlisle Ah, right, yes I see the logic

David Carlisle

@JosephWright meanwhile all checks in that PR now pass:-)

Joseph Wright

12:54 PM

@DavidCarlisle I have a solution to the $$ question ... should I post it?

David Carlisle

@JosephWright from the math grabbing? I don't know I sort of like sticking to the not supported line, but yes why not

egreg

@JosephWright No, if it allows using $$. ;-)

Joseph Wright

@DavidCarlisle Yes, exactly - we've had to come up with something ...

@egreg Well yes, and that was my original idea for tagging, but Frank is right that there at least realism has to win

\cs_gset_protected:Npn \__math_grab_dollardollar:w % $$
  #1 $$
  {
    \tl_if_blank:nF {#1}
      {
        \__math_process:nn { equation* } {#1}
        \socket_use:n {tagsupport/math/display/begin}
        \socket_use:nn{tagsupport/math/display/formula/begin}
          { \leftline{$\indent\displaystyle{#1}$} }
        $$
      }
  }

with tagging turned on ...

It looks horrible, BTW

(The output)

The question does in a sense sound like its in the same space as tagging - as it's about visual impairment and avoiding document changes

Ulrike Fischer

@JosephWright which they probably should do if they produce documents for visually impaired persons (not that I understand why left aligned math helps much).

Joseph Wright

@UlrikeFischer Yeah, that I don't follow fully - perhaps start-of-line is important and lost with centering?

Ulrike Fischer

1:01 PM

seing my upvotes it looks as if someone does some research about form fields ...

4

egreg

1:17 PM

@JosephWright Use the code to show “ERROR: $$ cannot be used if you want full support of tagging and everything else in LaTeX.”

3

gusbrs

@PauloCereda You did not! :-O That's a whole new level of "coding for fun". :-)

Paulo Cereda

@gusbrs I cheated. :) Edit: github.com/Annihil/github-spray

David Carlisle

1:35 PM

@JosephWright post an answer I suppose.

Joseph Wright

1:46 PM

@egreg Except we are tagging $$....$$ ;)

(Interesting that DEK more-or-less is saying don't use $$ in the trick from The TeXbook: I continue to want to drop it entirely and use only $\displaystyle ...$ , cf. breqn and ConTeXt)

gusbrs

@PauloCereda Quack!!!

mickep

@PauloCereda Oh, secret messages: tug.org/pipermail/tex-live/2024-March/050214.html

David Carlisle

@mickep Y2FuJ3QgeW91ciByZWFkIGJhc2UgNjQ/

Paulo Cereda

@mickep vvv

==========================================================================
Ubuntu Security Notice USN-6695-1
March 14, 2024

texlive-bin vulnerabilities
==========================================================================

A security issue affects these releases of Ubuntu and its derivatives:

- Ubuntu 23.10
- Ubuntu 22.04 LTS
- Ubuntu 20.04 LTS

Summary:

Several security issues were fixed in TeX Live.

Software Description:
- texlive-bin: Binaries for TeX Live

Details:

It was discovered that TeX Live incorrectly handled certain memory

@mickep we ducks are very good at secret messages :)

mickep

@DavidCarlisle Ohhh

@PauloCereda Too curious for your own best.

@PauloCereda I'm in China now. People here eat guys like you to breakfast.

Paulo Cereda

2:00 PM

@mickep oh no

David Carlisle

2:11 PM

@PauloCereda but can you read my comments?

@mickep doesn't everyone?

Paulo Cereda

@DavidCarlisle ooh

Ulrike Fischer

@DavidCarlisle how do one decode them?

David Carlisle

@UlrikeFischer (a) use @PauloCereda's favourite editor (b) run base64-decode-region

@UlrikeFischer see in the scratch buffer:

(base64-encode-string " how do one decode them?")
"IGhvdyBkbyBvbmUgZGVjb2RlIHRoZW0/"

Ulrike Fischer

@DavidCarlisle or find a suitable internet site ;-)

David Carlisle

@UlrikeFischer you are so modern

@UlrikeFischer but the similar decode-hex string makes it easy to read uncompressed pdf with hex utf-16 inline as you can decode the hex (meaning the pdf is invalid but you can always undo, or not save

Ulrike Fischer

2:31 PM

@DavidCarlisle ;-). But we are probably the only one who read a pdf directly, all the other use some clicking tools which decodes the stuff (or not, Peter complained about our encoding of the title and then it was its tool which had messed it up).

Joseph Wright

@UlrikeFischer, @DavidCarlisle Speaking of hex utf-16, lots of changes on updating tagpdf are about that :)

Ulrike Fischer

@JosephWright that is less tagpdf more the pdfmanagement as I added there the /Contents entry.

Joseph Wright

@UlrikeFischer Ah: I just wrote the shorter thing to say!

David Carlisle

3:33 PM

@JosephWright

! Use of \??? doesn't match its definition.
<argument> \???
                 ! LaTeX Error: Erroneous variable \l__socket_tagsupport/mat...
l.21 $$ double~dollars $$

?

@UlrikeFischer ^

Joseph Wright

@DavidCarlisle \listfiles?

David Carlisle

ah local file... :-)

@JosephWright no need:-)

hmm no still get it

Ulrike Fischer

@DavidCarlisle Joseph's example?

David Carlisle

@UlrikeFischer yes

I had a localy installed expl3 which I zapped and am rebuilding formats, will try again in a bit

Ulrike Fischer

@DavidCarlisle it fails for me in miktex, so something is missing ...

David Carlisle

3:43 PM

 *File List*
documentmetadata-support.ltx
pdfmanagement-testphase.sty    2024-03-01 v0.96f LaTeX PDF management testphase
 bundle
tagpdf-base.sty    2024-02-29 v0.98x part of tagpdf - provide base, no-op versi
ons of the user commands
pdfmanagement-testphase.ltx    2024-03-01 v0.96f PDF management code (testphase
)
l3backend-pdftex.def    2024-02-20 L3 backend support: PDF output (pdfTeX)
l3backend-testphase-pdftex.def    2024-03-01 LaTeX PDF management testphase bun
dle backend support:PDFoutput(pdfTeX)

@JosephWright ^^^

@UlrikeFischer @JosephWright works with pdflatex-dev

Joseph Wright

@DavidCarlisle Well yes, what else would you be using?

Ulrike Fischer

@DavidCarlisle yes, just realized that too.

David Carlisle

@JosephWright you might have "reminded" people in the answer:-)

Joseph Wright

@DavidCarlisle Slipped my mind - I routinely run the dev format

Paulo Cereda

@JosephWright % arara: <engine>: { branch: developer }

:)

David Carlisle

4:00 PM

@JosephWright thinking about a test suite for l3sys-query is there an easy way to run it in a known clean directory, I think normallly l3build will run all tests together or just one if that is specified, which leaves the directory contents a bit variable

Joseph Wright

@DavidCarlisle I'd just set up a test config with only one test ;)

David Carlisle

@JosephWright oh OK I had thought of that, but wondered if I'd missed something, I may do that

Joseph Wright

@DavidCarlisle Performance-wise, cleaning up after every test would slow stuff down

David Carlisle

@JosephWright sure, In general but I was wondering about a command to clean the working directory from l3build but as you say just using a separate config is more or less equivalent or current plan is just to not check the "tests provided" check box as our code is naturally bug free

Paulo Cereda

@DavidCarlisle % arara: clean: { extensions: [ ..., ..., ... ] }

:)

David Carlisle

4:12 PM

:65341844

Command 'arara' not found, did you mean:
  command 'arara' from deb texlive-extra-utils (2021.20220204-1)
Try: sudo apt install <deb name>

@PauloCereda your challenge for today (which is soluble): explain why that command is not found having done a full texlive 2024 install

Paulo Cereda

@DavidCarlisle that is only true for mac users. :)

David Carlisle

@PauloCereda debian wsl

Paulo Cereda

@DavidCarlisle oh no

@DavidCarlisle what has Debian ever done for us

David Carlisle

@PauloCereda possibly you are assigning blame in the wrong place here. read the error message really carefully

Paulo Cereda

@DavidCarlisle oh now I see, sorry :)

We ducks are very good at reading error messages

David Carlisle

4:19 PM

@PauloCereda would have been easier base64 encoded chat.stackexchange.com/transcript/message/65341309#65341309

Paulo Cereda

@DavidCarlisle ooh

David Carlisle

7:29 PM

@JosephWright doesn't work:-)

Joseph Wright

7:42 PM

@DavidCarlisle Er, what happens?

David Carlisle

@JosephWright well it works as defined, not as I want:-) it's in tools so l3build check -c config-l3sys first builds base and tools before running the test and so the directory is rather full (and I ended up with a 1500 line log file)

Joseph Wright

@DavidCarlisle Selective ls, or we do some kind of Lua-based testing that ... I don't know about just yet ;)

David Carlisle

@JosephWright yes I could just do more specific tests like looking for someprefix* (which is probbaly enough really) although it would be nice to test . or * for example, I'm having a think now, I may be able to do a custom checkinit function...

Joseph Wright

@DavidCarlisle Ah, that would work

@DavidCarlisle I was imagining you'd do ls on the test directory

David Carlisle

7:57 PM

@JosephWright l3sys-query pwd reported I was in PWD: /home/latex3/latex2e/build/test-config-l3sys is that what you meant?

Joseph Wright

@DavidCarlisle Yes, should not be that busy

David Carlisle

@JosephWright er...

(|l3sys-query ls '*.tex')
star: ./chapter1.tex
star: ./chapter2.tex
star: ./chaptera.tex
star: ./checkencodingsubset.tex
star: ./docstrip.tex
star: ./e.tex
star: ./fontsmpl.tex
star: ./foo^^7dbar.tex
star: ./glyphtounicode-cmex.tex
star: ./glyphtounicode.tex
star: ./h.tex
star: ./idx.tex
star: ./lablst.tex
star: ./lcydefs.tex
star: ./lipsum.ltd.tex
star: ./lppl.tex
star: ./ltluatex.tex
star: ./ltxcheck.tex
star: ./nfssfont.tex
star: ./q.tex
star: ./r.tex
star: ./regression-test.tex
star: ./s.tex

Joseph Wright

@DavidCarlisle Like I said, not that busy

David Carlisle

@JosephWright oh hang on, * gave that but...

@JosephWright why in the same run did I get this

Joseph Wright

Meanwhile, I have some light reading: Christoph Burnikel and Joachim Ziegler. Fast Recursive Division. Research Report MPI-I-98-1-022, Max-Planck-Institut f\"ur Informatik, Im Stadtwald, D-66123 Saarbr\"ucken, Germany, October 1998

David Carlisle

8:01 PM

(|l3sys-query ls --type f )
f _: ./afterpage.sty
f _: ./alltt.sty
f _: ./amsbsy.sty
f _: ./amscd.sty
f _: ./amsgen.sty
f _: ./amsmath-....-..-...sty
f _: ./amsmath.sty
f _: ./amsopn.sty
f _: ./amstex.sty
f _: ./amstext.sty
f _: ./amsxtra.sty
f _: ./ansinew.def
f _: ./applemac.def
f _: ./array-....-..-...sty
f _: ./array-....-..-...sty
f _: ./array-....-..-...sty
f _: ./array.sty
f _: ./article.cls
f _: ./article.sty
f _: ./ascii.def
f _: ./ascii.tcx
f _: ./atbegshi-ltx.sty
f _: ./atveryend-ltx.sty

Joseph Wright

I suspect @DavidCarlisle might be familiar with it

@DavidCarlisle Did you start from a clean directory? (I do think we might need custom code here as the likelihood is 'stuff' being about)

@DavidCarlisle How about set up to copy some 'support' files into a subdir, then you do control what is present

We have something like that for file name testing

David Carlisle

@JosephWright I started with a custom support dir full of test files to list, and it's building in a new build/test-config-l3sys directory but as I say it's building base first. I'll look, as I say I think I should be able to force a known state in the check init hook

Joseph Wright

@DavidCarlisle Yes, I think the best plan

David Carlisle

@JosephWright just checked directly `/home/latex3/latex2e/build/test-config-l3sys
` has 426 files in it

Joseph Wright

@DavidCarlisle Still not too long in .tlg terms

David Carlisle

8:10 PM

@JosephWright yes but Its very fragile, as for example if you do the option not to build the dependencies first it doesn't build base and then 400 of those 426 files aren't there and the test fails...

Joseph Wright

@DavidCarlisle Right yes - so sticking to a setup approach likely best - make a subdir and use that I think

David Carlisle

@JosephWright yes

Joseph Wright

9:00 PM

@DavidCarlisle Working?

David Carlisle

@JosephWright too well... l know what to do but left it for now. but as general advice it's not a good idea to work in a clean directory and turn off searching the main tex tree:-)

Joseph Wright

@DavidCarlisle :)

@DavidCarlisle I realise that the PDF docs for l3sys-query got missed out, because of the rename: fixing for an update tomorrow morning :)

yo'

10:32 PM

Seems the drama is over.

3

Joseph Wright

11:10 PM

2

Transcript for

$TeX - LaTeX$ TeX, LaTeX and Friends