« first day (3269 days earlier)      last day (30 days later) » 

7:22 AM
@AlanMunn, @schtandard Hmm .. OK ...
7:46 AM
@DavidCarlisle I'll sort PL1 this morning I guess: we are OK to go?
@DavidCarlisle Also need to tidy up some branches
@egreg Hope the doc update is OK
I see that the latest L3 update is already in TeX Live
@StefanKottwitz Everything looking good with tex.ac.uk? The registration got moved from JISC to a company, and I'm now formally the contact
@AlanMunn, @schtandard I'd strongly recommend the Atlassian Git tutorial series: atlassian.com/git/tutorials/learn-git-with-bitbucket-cloud
8:06 AM
@JosephWright Yes I think graphics will need another update but can go now with base
@DavidCarlisle OK, I'll sort out that one merge issue and build for release
@DavidCarlisle Meanwhile, see 'LaTeX2020' (fingers-crossed I got all of the tests right)
@JosephWright did I got it right that one can now build the dev format with the ctan l3kernel?
Good morning to all users into chat.
Please, can I ask an help for google chrome if it possible (out the LaTeX matter)?
8:22 AM
@UlrikeFischer Yes :)
@UlrikeFischer 'LaTeX2020' version now on CTAN, and I'll merge some stuff at the 2e end shortly
@JosephWright nice ;-).
@UlrikeFischer Just sorting the merge at the 2e end, then it will be in develop there; then I'll do the PL1 release
8:39 AM
@JosephWright I'm currently converting tagged to dtx and changing the module name to "tag" and I will start to add the pdfresource stuff. But I don't know if I can do much today - we are driving to Palermo.
@UlrikeFischer Sounds good
@UlrikeFischer I've got some 'real' work to do for tomorrow, so mainly am going to have my PC grinding away for PL1 (deployment is not on master)
@JosephWright today I'm one year in the team, so I put on my shirt - if we find a suitable place we will make a foto ;-)
@UlrikeFischer YAY!!!
@UlrikeFischer I like Norbert's comment:
> Because I didn't think of that and the check is not in the code ;-) I
actually never thought anybody will disable the create_formats options,
it was there due to historical and portable reasons.
@JosephWright yes, I saw the thread yesterday. I wonder why the missing format update didn't pop up more often.
@UlrikeFischer Yes, this is very odd
@UlrikeFischer We need it fixing for ... reasons
8:54 AM
@JosephWright secret reasons
@UlrikeFischer Well yes
@UlrikeFischer I thought that TL had a list of packages that forced a format-update ... seems I'd totally misunderstood
@UlrikeFischer you have been in a year and there are still 583 unclosed tickets in gnats, we clearly haven't passed enough blame on to you!
@DavidCarlisle, @UlrikeFischer I'm working out a few issues with building docs ... the .err files are causing an issue on the LaTeX2020 branch. Once that is done, I'll get PL1 sorted
9:10 AM
@JosephWright so to delete graphics-compat here and remote it's something like git branch -d graphics-compat ; git push -d graphics-compat ? or is there a shorthand to do both at once? (also amsmath-2018, curr-file-braces and xcite can go I think)
@JosephWright YAY!!
@UlrikeFischer yay!
@DavidCarlisle yay!
Perl 6 = Raku
> Over 100 published studies may have incorrect results thanks to a glitchy piece of Python code discovered by researchers at the University of Hawaii. An anonymous reader quotes Motherboard: The glitch caused results of a common chemistry computation to vary depending on the operating system used, causing discrepancies among Mac, Windows, and Linux systems.
@DavidCarlisle ^^ not Ulrike's fault?
@PauloCereda are you sure it is not?
ooh a plot
@PauloCereda 25 to go
@DavidCarlisle Some GUIs can do it, but otherwise it's a two-step, yes
@DavidCarlisle I can tidy up if you like
9:21 AM
@JosephWright my mother always taught to me to tidy up after myself (not that I always obeyed:-)
@JosephWright GK master. :)
@DavidCarlisle ooh
@DavidCarlisle Tidied up at the remote
@DavidCarlisle Building PL1 on my VM: it's a lot faster than native Windows. Will do that, then send to CTAN, then work out why docs are not building on LaTeX2020 (seems to relate to the .err files ...)
Hi, latex-dev is not updated after the official release of latex; is this correct? The enthusiastic testers might be still using latex-dev 2019-10-01-pre-release, even though latex 2019-10-01 is already released.
@HironobuYAMASHITA No, we've not updated just yet: that's still something we have to think about (a PR1 that's just the same as the main release is a bit ... odd)
@HironobuYAMASHITA We have to watch the load on CTAN too
@HironobuYAMASHITA the "real" next latex-dev will be off the latex2020 branch (see joseph's comment above) I think we need to still work out a good way to update main and dev without over-stressing the ctan maintainers
9:34 AM
@DavidCarlisle Probably next time we should do a 'gold' dev release that will be the same as the new master, say with a week of 'grace'
@JosephWright well either that or (perhaps simpler) push the dev branch on before updating main.
@DavidCarlisle Hmm, could be: I mean that we likely need to do two releases close together, at least of base, whether we like it or not
@DavidCarlisle Speaking of 2020, I think I have the test suite passing, just have to track down a problem building the docs (I do like that Travis-CI can do that, so I know that a release can go straight out)
I have Travis-CI building one test of tests (2020), my VM building master, and my 'real' machine building the dev branch :)
Now, how do I convince someone to buy me a MacBook Pro i7 ...
@JosephWright @PauloCereda buy @joseph a new macbook
@JosephWright let's see if ^^ works
@DavidCarlisle I was thinking more @egreg ;)
@JosephWright In a few years…
9:39 AM
@egreg :)
@DavidCarlisle I could ask UK-TUG ;)
@JosephWright skype
@DavidCarlisle OK
@JosephWright bah I was going to suggest this :)
@JosephWright I want one too
@DavidCarlisle I deny any blame for python - too many spaces.
@UlrikeFischer and indentation. :)
Also, snakes are dangerous to bears and marmots. :)
9:48 AM
@JosephWright sure
10:09 AM
anyone want to try this with luatex:


for i,j in pairs(texconfig) do
print (i .. ': ' .. v)
@JosephWright @PauloCereda ^^
what does the luatex manual mean by "A startup Lua script could fill this table with a number of settings that are read out by the executable after loading and executing the startup file." where can I shell_escape_commands list form if it is not in that table?
This is LuaTeX, Version 1.10.0 (TeX Live 2019)
 restricted system commands enabled.
LaTeX2e <2019-10-01>

luaotfload | main : initialization completed in 0.252 seconds
Document Class: article 2019/08/27 v1.4j Standard LaTeX document class
No file test.aux.
kpse_init: 2018
 384 words of node memory still in use:
   2 hlist, 1 vlist, 1 rule, 2 glue, 3 kern, 1 glyph, 5 attribute, 44 glue_spec
@DavidCarlisle ^^
@PauloCereda yes so how does that fit with section 10.4 of the manual saying kpse_init is a boolean? and where are the other values...
@DavidCarlisle let me check
10:29 AM
@DavidCarlisle Ask @MarcelKrüger
@UlrikeFischer this was supposed to be an easy "close an open issue" activity, just logging \ShellEscape not opening a can of worms in luatex internals:(
11:08 AM
@UlrikeFischer @PauloCereda (@MarcelKrüger) found it, you can use kpse library instead:
@UlrikeFischer awwww <3
@campa Yep, I think this is it! Cheers
11:27 AM
@UlrikeFischer guess what we are having for lunch
local se_commands = "," .. kpse.var_value('shell_escape_commands') .. ","
function se_is_allowed (c)
if(string.find(se_commands, "," .. c .. ",",1,true)) then
print(c .. ": yes")
print(c .. ": no")


@DavidCarlisle Do you need this only when shell escape should be invoked for logging purposes or do you also want to have this information independently?
@MarcelKrüger currently just logging (and just for \ShellEscape/\write18 not arbitrary os.execute calls in Lua code, see github.com/latex3/latex2e/issues/195 I can implement something now, or are you wondering if luatex could optionally log this in its internal code?
@DavidCarlisle The easier way is to look at os.executes etc. return values. The first return value is nil iff the command wasn't allowed. Then the second return value contains a reason. Otherwise the first return value will be a number:
  local status, msg = os.execute("some crazy command")
  if status == nil then
    % Command wasn't allowed
  elseif status == 0 then
    % Command executed successfully
    % Command has been executed but failed
    print("Something went wrong. Command exited with status " .. status)
@MarcelKrüger ah of course that is much safer than me trying to re-do the test in lua "by hand" thanks!
1 hour later…
12:37 PM
@JosephWright Yes!
1 hour later…
1:48 PM
PS: sorry, the previous link was to a mass in the national shrine yesterday... I just sent the link to my dad :)
@PauloCereda Do you know what's the background music?
@egreg oh... no
@PauloCereda “Il trillo del diavolo” by Giuseppe Tartini: en.wikipedia.org/wiki/Violin_Sonata_in_G_minor_(Tartini)
@egreg ooh!
2:05 PM
@PauloCereda Tartini was “maestro di cappella” for the Basilica del Santo.
@egreg how nice!
@PauloCereda His tomb is in a small church in Padova: it.wikipedia.org/wiki/Chiesa_di_Santa_Caterina_(Padova)
@egreg you do know this church, right? :)
@PauloCereda The monastery next to it now is the Department of Statistics.
@egreg oh no :)
Statistics is dangerous
2:10 PM
@PauloCereda It used to be an orphanage, after monastic properties were seized in the Napoleonic era.
@egreg ooh
@AlanMunn: toopanda.com/…
2:36 PM
user image
@JosephWright @DavidCarlisle @egreg ^^^ Anniversary foto "One year in the team" ;-)
@UlrikeFischer yaaaaaay
2:47 PM
@PauloCereda diabolic!
Oh great, I think I've found a bug in pdfTeX ...
@JosephWright ?
Or is it ... a bug in TeX???!!! That cannot be
@JosephWright ?
@UlrikeFischer Build a format that does \everyjob{\edef\demo{\jobname}} and then try looking at \jobname when you use it ...
@DavidCarlisle ^^^
\catcode`\{=1 \catcode`\}=2 %
2:59 PM
@JosephWright is this in 2020?
@DavidCarlisle Yes: tracking down the doc-building failure
@DavidCarlisle Comes down to the same thing: trying to use \jobname inside \everyjob
@PauloCereda You have entirely too much time on your hands... Don't you have a thesis to finish?
@UlrikeFischer, @DavidCarlisle, @egreg Could you check my test?
@JosephWright vv
> \temp=macro:
<*> \show\temp


> \test=macro:
<*> \edef\test{\jobname}\show\test

@DavidCarlisle Yes, that's the one
@DavidCarlisle I can patch round to some extent for the present, but this seems ... odd
@DavidCarlisle Was that pdfTeX or TeX90?
3:03 PM
@JosephWright actually that was at * so texput is expected, if I put it in a file I get
pdftex \&zf zf2
This is pdfTeX, Version 3.14159265-2.6-1.40.20 (TeX Live 2019) (preloaded format=zf)
restricted \write18 enabled.
entering extended mode
> \temp=macro:
l.1 \show\temp

> \test=macro:
l.2 \edef\test{\jobname}\show\test

Hi fellow stackers. I was just wondering how you guys look back at the site design change to the more "responsive design". I saw a lot of people threatening to leave or even to set up their own community at that time. How did that work out? Obviously the site is not completely dead :)
@DavidCarlisle OK, so I'm doing something wrong??
I hope I'm not delving too deep in an open wound here...
@JosephWright or its in the win32 specific bits? (that was cygwin)
@MaartenBodewes personally I use a user css file that restyles it to something I found more usable
@DavidCarlisle Ah, got it ...
@DavidCarlisle It's about exactly how you call the binary
3:05 PM
@MaartenBodewes I think most people just stayed around and put up with it. Some people added custom CSS to fix some of the awful parts. A couple of people might have left, but people leave for lots of other reasons anyway, so it's hard to tell if anyone specifically left due to the design changes.
@DavidCarlisle You did \&<fmtname>, I did -fmt=<name>.fmt. If you remove the extension, it works ....
@DavidCarlisle Phew, I can fix at the build script end
@AlanMunn Sounds like the same situation as it was at cryptography. However, if possible, the site users here were even more vocal, so I was rather worried about this community!
@DavidCarlisle the same as I get. Both Knuth TeX and pdftex
@JosephWright vv
$ pdftex -fmt zf.fmt zf2
This is pdfTeX, Version 3.14159265-2.6-1.40.20 (TeX Live 2019) (preloaded format=zf.fmt)
 restricted \write18 enabled.
entering extended mode
> \temp=macro:
l.1 \show\temp

> \test=macro:
l.2 \edef\test{\jobname}\show\test

We already had a rather modern looking site, so it didn't affect us as much as $\TeX$ or IT Security.
Oh, no TeX here of course :)
3:07 PM
@JosephWright did you do this:
$ pdftex -fmt zf.fmt \\input zf2
This is pdfTeX, Version 3.14159265-2.6-1.40.20 (TeX Live 2019) (preloaded format=zf.fmt)
 restricted \write18 enabled.
entering extended mode
> \temp=macro:
l.1 \show\temp

> \test=macro:
l.2 \edef\test{\jobname}\show\test

? x
which is same as
@JosephWright I did pdftex --fmt=testformat.fmt blub1 and get for both \show commands blub1
$ pdftex \&zf  \\input zf2
This is pdfTeX, Version 3.14159265-2.6-1.40.20 (TeX Live 2019) (preloaded format=zf)
 restricted \write18 enabled.
entering extended mode
> \temp=macro:
l.1 \show\temp

> \test=macro:
@MaartenBodewes Yes, those of us who had been here from very early on spent quite a bit of time with the original site design, and since typography is something that lots of TeX users care about, the changes certainly hit a nerve with some. And the way SE "sold" the change was very disingenuous, which also annoyed lots of people.
@UlrikeFischer TL or MiKTeX?
@DavidCarlisle Ah ...
@JosephWright TL.
3:09 PM
@DavidCarlisle I'm probably being thick
@MaartenBodewes But ultimately the site still serves its main purpose, and this is a great community overall, so most people stayed.
@DavidCarlisle Thing is, this works fine if you don't set \everyjob, so there's something not-quite right
@DavidCarlisle, @UlrikeFischer I'll write something to the TL list with a logical story: a bit busy now, so later today
@AlanMunn Great it survived, and I'm very happy that it exists, I always use it as my first reference when looking for crypto related issues (and since crypto is pretty broad in the math etc. it uses, that covers quite a lot in itself).
@JosephWright i think \jobname is set if you use it so normally it gets set when you reference the file, or at the \input but here it gets used in everyjob before the \input
Separately, if anybody of you have got remarks on our crypto specific formulas at our Meta, don't hesitate to contact me or our user fgrieu in our own chat!
A: Using MathJax / $\TeX$ on the Cryptography site

Maarten BodewesMathJax is used to format mathematical expressions and symbols on cryptography site. You can use expressions within single dollar signs for inline math: $0101_2$ which gives and $0101_2$. Use double dollar signs for a centered block instead: $$\mathtt{CAFE}_{16}$$ gives $$\mathtt{CAFE}_{16}$$ I...

3:13 PM
@DavidCarlisle Right, but the weird thing is it changes \jobname, and that's supposed to be impossible from the macro layer
@JosephWright jobname starts as texput so it's not so much changing it as freezing it before you know the filename.
@DavidCarlisle OK, so like I said, this is fixable in our script for the present (by not using \input), but is at least something deserving of comment from Karl et al.
@MaartenBodewes I'm not a mathematician, so would have nothing to add here, but that answer looks like a very useful reference sheet.
@MaartenBodewes you might (or might not) want to mention that mathjax specific questions are off topic here search or post a question on the $\TeX$ sister site; so if asking here users should try to frame it as a tex question
@DavidCarlisle That's a very good point!
3:17 PM
@MaartenBodewes mathjax specific questions can be asked on mathjax tag at stackoverflow
@DavidCarlisle Do you think the behaviour is predictable/expected?
Ah, right! Although I don't see many MathJax related issues fortunately. Generally it is just the problem that people don't know it exists in the first place.
@JosephWright it's probably explainable given documented statements.
@DavidCarlisle Ah, I think I addressed this for tests: it's just typesetting that has the issue. I'll adjust in l3build, and for the present in the config for LaTeX20202
@JosephWright Well I think if you do \edef\temp{\jobname} \jobname must have a meaning and so freezing it sounds sensible.
3:23 PM
@JosephWright you get same with
\catcode`\{=1 \catcode`\}=2 %
@JosephWright jobname is the filename if that is on the commandline, but texput if you do \input file
OK, added additional references and made sure no MathJax questions get asked here. Thanks already!
@JosephWright regarding tests: I tried in latex2e\required\graphics to run l3build check -eluatex tlb0610 and it fails with the message lualatexquotejobname.lua not found. Is that expected?
I'll let you get on with the building / programming.
@MaartenBodewes normally chat around here is about ducks and cricket, you just happened to wander in on a day we updated the public latex release, so chat a little more tex related than usual:-)
Right. As a very frequent kayaker I'll get back here sometimes and scare the ducks away :P
3:34 PM
@PauloCereda ^^
Actually have to pick up my kayak right now for a rolling session in the pool we rent. Don't worry, no ducks there (I sincerely hope). TTYL!
2 hours later…
5:51 PM
@UlrikeFischer is there any miktex question for which that isn't the answer?
2 hours later…
7:32 PM
@UlrikeFischer Er, I'll have to look at that
7:46 PM
@DavidCarlisle Is the #195 fix done? To merge to dev?
7:56 PM
@DavidCarlisle Probably not ;-)
8:23 PM
Good evening and good sunday.
8:45 PM
@JosephWright yes, I was just making sure you'd finished pushing out the release before changing it again, I can merge it over now.
@JosephWright there were a couple of design issues (like exposing the number as \ShellEscapeStatus which I slipped in, but I think it can go to master anyway.
@JosephWright This shows up as expl3 needs to use \jobname in the \everyjob hook. that may be as issue as we found when doing the first utf8 changes, several IDE including emacs auctex always use the \input form so are they going to get jobname=texput?
9:03 PM
@DavidCarlisle No one noticed the pearl in ulem.sty, which defines its own version of \MakeRobust. This breaks the accent redefinition by amsmath (when they are made robust by the last release).
@egreg I mailed Donald the other day to ask for a change to url.sty:-) need to do it again?
@DavidCarlisle Yes: \MakeRobust has been a kernel command for a few years!
@egreg is this from that wiley class question? I didn't have time to unpack the zip file to see what it was doing
9:20 PM
@DavidCarlisle Yes. Here's the MWE



@egreg yes just saw your answer (voted even:-)
@DavidCarlisle It works if stix is loaded after amsmath. But the \MakeRobust problem remains.
@DavidCarlisle You'll be sad knowing that I was waiting for rep cap and your vote provided the 10 points.
@egreg and it would work if ulem did \ifx\MakeRobust\undefined \def\MakeRobust...\fi it would work?
@DavidCarlisle Well, that's for people using quite an old kernel. But it could be better anyhow, so rollback is not a problem
@egreg ulem written to work with plain as well
9:24 PM
@DavidCarlisle In case of plain, \MakeRobust is defined to do nothing.
@egreg yes I know :-)
9:40 PM
@DavidCarlisle I've just fixed the issue with \includegraphics in tex4ht. All what was needed was to use one \edef and \detokenize.
@michal.h21 thanks I was going to say that it was likely to be catcodes of the filenames
@DavidCarlisle yes. I was just trying to fix \Gin@ext, but the issue was with a temporary macro in graphics.4ht
10:24 PM
@DavidCarlisle Yes: I'll write something to the team list tomorrow

« first day (3269 days earlier)      last day (30 days later) »