12:22 AM
That seems to work, if you don't write the answer I'll try to write one up later
Have to go sterilize a lab now

3 hours later…
3:49 AM
Anyone seeing lots of pdf backend warnings about duplicate destinations with the latest nicematrix update with lualatex?

4 hours later…
7:37 AM
@LaTeXereXeTaL can you show a small complete example?

8:09 AM
@UlrikeFischer did the generic bit of xkeyval get into miktex properly? (see recent question)

@DavidCarlisle I'll check.
@DavidCarlisle the generic files are missing. I will open an issue.

@UlrikeFischer thanks it looked that way from the OPs log :-)

8:25 AM
@DavidCarlisle hm, the texlive installation isn't faultless either. They seem to have put the examples into tex/generic.

@UlrikeFischer sometimes I wonder about the ctan preference for not having tds archives, and leaving it up to miktex and texlive to guess where to install stuff

@DavidCarlisle yes, it certainly opens place for errors, not only for the tex systems but also for users installing manually.

2 hours later…
10:05 AM
> The Best Nintendo Switch Games To Avoid Uncomfortable Interactions With Distant Relatives on Thanksgiving
@samcarter_preparing_for_xmas ^^ you gotta love some headlines. :)

@PauloCereda :D I thought that's why we do all this covid nonsense :)
Luckily my desk lamp broke yesterday, so I had to make a trip to the hardware store and there was this cute duck waiting to be adopted:

@samcarter_preparing_for_xmas :)
@samcarter_preparing_for_xmas OH MY <3

@PauloCereda One must simply love the colour!

@samcarter_preparing_for_xmas indeed!

10:35 AM
@DavidCarlisle do you know how (in)compatible tabulary is with arydshln?

10:50 AM
@UlrikeFischer I may have done once but not now. I can't remember is arydshln compatible with array package at all? If it is it should work with tabularx/y as that's just a normal tabular* after it has guessed the width to make p columns.

\documentclass{article}

\usepackage{tabulary}
\usepackage{arydshln}

\begin{document}
\begin{tabulary}{10cm}{CCC}
& Mo & Di\\
10:00 & Deutsch & Englisch\\
11:00 & Deutsch & Kunst\\
\end{tabulary}
\end{document}
\documentclass{article}

\usepackage{array}
\usepackage{arydshln}
\usepackage{tabulary}

\begin{document}
\begin{tabular}{lll}\hline
A&B&C\\\hdashline
\end{tabular}

\begin{tabulary}{10cm}{CCC}
& Mo & Di\\%\hdashline
10:00 & Deutsch & Englisch\\
11:00 & Deutsch & Kunst\\
\end{tabulary}
\end{document}
gives
! Missing # inserted in alignment preamble.
&
l.31 \end{tabulary}
So I think I will say "you will have to choose ..."

Good morning all, I want to get rid of pkg pdftexcmds, but I cannot get the L3 replacement working. For example pdffiledump. This is my example:
\documentclass{article}

\usepackage{pdftexcmds}

\begin{document}
\ExplSyntaxOn
% works
\lua_now:e{
oberdiek.pdftexcmds.filedump( "0", "8", "\lua_escape:e{\jobname.tex}" )
}
~
% doesn't
\lua_now:e{
ltx.util.filedump( "\lua_escape:e{\jobname.tex}", "0", "8" )
}
\ExplSyntaxOff
\end{document}

@AlexG Where is ltx.util from? The expl3 equivalent would be l3kernel.filedump

@AlexG there is an s missing in utils, and I think the numbers should be numbers. And you need to do something to print it,
\lua_now:e{
tex.sprint(ltx.utils.filedump( "\lua_escape:e{\jobname.tex}", 0, 8 ))
}

@PhelypeOleinik I took it from texdoc interface3, page 257.

11:06 AM
@AlexG Oh, l3kernel is deprecated. Sorry

@PhelypeOleinik ooh l4kernel :)
2

@PauloCereda Only 2000 years to go

@UlrikeFischer Oh, a typo!
@UlrikeFischer Thank you, this works!!!
( I am luatex almost illiterate.)

@UlrikeFischer I suppose I should look... especially for horizontal dashes it must be trivial to make work as they are self contained, vertical dashes have a larger impact on the whole table buildup.

@UlrikeFischer We should probably add tonumber() here

11:15 AM
@PhelypeOleinik In the year 2525... :)

@PauloCereda ... if ducks are still alive

@DavidCarlisle oh no
@DavidCarlisle do you know Futurama's version of this song? :)

@AlexG The “official” name for \jobname is \c_sys_jobname_str

@PauloCereda probably not. Sadly though non-biodegradable plastic ducks may still be here in 2525

11:25 AM
@DavidCarlisle woo

@PauloCereda wrong response to an environmental damage comment

@DavidCarlisle oh

@DavidCarlisle I think I can obsolete textcase :)

@egreg Oh, didn't know that. Thank you!

@DavidCarlisle That's would only be a problem, if people would throw away plastic ducks - which of course nobody does because one keeps such nice things

11:34 AM
@samcarter_preparing_for_xmas exactly

@JosephWright there are only 60 lines of code to replace:-)

@DavidCarlisle See Skype
@DavidCarlisle Sure :)
@DavidCarlisle I do worry that the expandable code isn't efficient, but it's as close-to as short as I could make it (if we need more speed, I can drop some bool operations and do things old-style)
@DavidCarlisle Did you see my email?

@JosephWright er, no, let me catch up..

@PauloCereda ketchup is red

11:40 AM
@DavidCarlisle ooh

11:52 AM
Regarding f-type expansion \romannumeral-0. Is it known who actually discovered this trick?

@HenriMenke I'd guess Heiko or Alan Jeffrey, or of course DEK himself, but I can't be sure: it certainly goes back a long way (it's been in expl3 more-or-less for ever, so late 1980s or early 1990s)

@JosephWright Is there version control information from these days to see who checked it in?

@PhelypeOleinik is this supposed to loop?
\documentclass{article}
\ExplSyntaxOn
\NewDocumentCommand\test{m}{#1}
\NewCommandCopy\origtest\test
\RenewDocumentCommand{\test}{m}{something\origtest{#1}}
\ExplSyntaxOff

\begin{document}
\origtest{abc}
\end{document}

@PauloCereda ^^^

@HenriMenke Er, no, I don't think even the 'private' SVN goes back that far (plus the older checkins were done by Rainer 'on behalf' of others, so the commit messages don't really help)

11:58 AM
@JosephWright Too bad, so I'll keep calling it “romannumeral trick”.

@HenriMenke What I do

@UlrikeFischer, @JosephWright Just discovered \file_hex_dump:nnn. But it doesn't seem to work, as it returns nothing:
\documentclass{article}

\begin{document}
\ExplSyntaxOn
\lua_now:e{
tex.sprint(ltx.utils.filedump( "\lua_escape:e{\c_sys_jobname_str.tex}", 0, 8 ))
}
\par
% (empty)
\file_hex_dump:nnn{\c_sys_jobname_str.tex}{0}{8}
\ExplSyntaxOff
\end{document}

@HenriMenke LOL! in my mind they will be microcrisps hereafter. :p

@AlexG Hmm, we have some index issues here: you need to start from 1 in the TeX code, but it looks like the Lua code starts from 0. We've standardised on index-from-1, but that's probably fuzzy when emulating primitives that index from zero. I'll log an issue

@HenriMenke LOL

12:07 PM
@JosephWright Thanks!

@AlexG github.com/latex3/latex3/issues/825 - I want to see if my analysis is right before adjusting the code
@AlexG I think pdfTeX indexes from 1 in any case, so you should be saying {1}{8} not {0}{8}. But I'd like some testing (@UlrikeFischer, @DavidCarlisle, @MarcelKrüger, @PhelypeOleinik could all look as well as me)

find of the day: \mathrm{\exp}, hmmm

@HenriMenke possibly David K (he certainly popularised it)
@JosephWright ^

@UlrikeFischer I can do better. I sent the .log to the maintainer and while I was sleeping the replied to confirmed a bug, sent a fix which works, and said the fix wold soon be pushed to CTAN.

@JosephWright pdftex manual says: ... must be >=0

12:23 PM
@JosephWright looks like 0 8 is correct to get the first 8 bytes according to th emanual

@JosephWright , @DavidCarlisle I ran a test, in pdftex, offset is zero-based. With offset=1 , the result is different:
\documentclass{article}

\begin{document}
\ExplSyntaxOn
%\lua_now:e{
%  tex.sprint(ltx.utils.filedump( "\lua_escape:e{\c_sys_jobname_str.tex}", 0, 8 ))
%}
\pdffiledump offset 0 length 8 {\c_sys_jobname_str.tex}
\par
\file_hex_dump:nnn{\c_sys_jobname_str.tex}{1}{8}
\ExplSyntaxOff
\end{document}
@JosephWright I find it unfortunate that \file_hex_dump:nnn wants an absolute index as second arg rather than a length (number of bytes) that pdftex and luatex expect.

@AlexG OK, so some adjustment is required
@AlexG I'll check hy we've done that

@JosephWright because REASONS? :)

@AlexG Ah, I remember
@AlexG In general, all of our 'range' functions take start-and-end, not start-and-length - for example \tl_range:nnn
@AlexG That's also why the TeX functions should index from 1: it's consistent with other parts of expl3

@JosephWright wait, 1-based indices?! That should be the 0-th item in your TODO. :)

12:35 PM
@JosephWright Makes sense. But as far as reading bytes from a file is concerned...

@JosephWright I was going to comment to @AlexG that the 0 offset is still consistent with the index being 1-based. The first byte is index 1.
@AlexG agreed but you can make your own user function that takes offset and length and calculates start and end so expl3 can do the reverse calculation to pass to pdftex.

@DavidCarlisle That's possible, of course.

@AlexG I see that it's n unusl case: I think we likely need to have a biref discussion on GitHub to work out how things should go
@AlexG Lu also indexes from 1 everywhere

@JosephWright don't make me hit you with my Flores book on computer arithmetic. :)
2
"Should array indices start at 0 or 1? My compromise of 0.5 was rejected without, I thought, proper consideration." — Stan Kelly-Bootle
@JosephWright ^^ :)

12:42 PM
@PauloCereda Futurama is awesome :D

@PhelypeOleinik <3

@PauloCereda given that I spend my day job converting between C and Fortran array indices are a big issue:-)

@UlrikeFischer I think so, because I didn't add \NewCommandCopy/\ShowCommand support to xparse (coming soon :)

@DavidCarlisle ooh :)

@PauloCereda ??

12:46 PM
@UlrikeFischer Yes, without xparse support, that is the same as \let\origtest\test, and then \origtest recurses to (the new) \test

@JosephWright computer arithmetic is a... fun subject. :)

@AlexG The main use cawses seem to be either 'all the file' or a predictable part, plus one can use expressions in those arguments to do a start pos. and an end-pos based on start + length

@DavidCarlisle :)

@PhelypeOleinik you mean if \test is an xparse command?

12:52 PM
@JosephWright Allow 0 as functionally-equivalent to zero did you mean 1 there?

@UlrikeFischer Yes, \NewCommandCopy has to “know” what an xparse command looks like to copy it, but it still doesn't

@JosephWright start, start+length-1 ;-)

@DavidCarlisle |E, yes

@PhelypeOleinik ah, somehow I thought it would already support xparse commands. But currently it is only \DeclareRobustCommands? How do one get the documentation with texdoc?

@JosephWright OK then I don't agree but at least I understand the suggestion:-)

1:00 PM
@UlrikeFischer \DeclareRobustCommand, \newcommand with optional argument, and (etoolbox) \newrobustcmd... Basically the same as \LetLtxMacro :)
@UlrikeFischer There isn't a documentation other than what's in ltnews or the source... sorry :)

1:12 PM
@PhelypeOleinik @DavidCarlisle will be pleased to hear ;-).

@UlrikeFischer :)
@UlrikeFischer I wonder where do we document these new stuff

@PhelypeOleinik the documentation of latex is a bit difficult to access anyway. You can get source2e with texdoc, but the rest is a bit of a chance game if you don't know the exact file name you are looking for.

1:50 PM
@JosephWright I'm confused by your last gh comment the version with % doesn't (empty) works and produces the same output as the line above doesn't it?

@DavidCarlisle Yes, I didn't remove @AlexG's comment line: guess I should

2:02 PM
@JosephWright ah I must say I have often posted on this site OP's mwe with "this doesn't work" comments after I'd made it work, so I should have spotted that:-)

OH NO

@PauloCereda LOL!

2:18 PM
@JosephWright @DavidCarlisle did you see the \romanumeral mail on the luatex list? Doing \romannumeral2000000000 \end with pdflatex is quite impressing too ;-)

@UlrikeFischer No??

ooh naughty romans
What have \romannumeral's ever done for us?

@JosephWright it segfaults with luatex. But with pdflatex it prints lots, really lots of "m". The romans seems not to have something for such large numbers ;-)
3

@UlrikeFischer LOL

2:31 PM
Just got some comments back from my thesis advisor and he specially mentioned that my tikz graphics are beautiful. :p
4

@Plergux ooh congrats!
:D

@PauloCereda :D thanks. :) I haven't told him yet that it's tikz (LaTeX). He probably thinks it's R graphics or Python or something. :p

@Plergux ooh :)

@PauloCereda but I'm probably going to tell him soon. seeing as how he's impressed :p
@PauloCereda When I mentioned it first in connection with my thesis he said something like "yeah, you should probably have learned that ten years ago." XD

@Plergux :)
@Plergux ooh :D

2:41 PM
@Plergux In the old days with thesis set on typewriter you had to do good mathematics to impress your supervisor, now all you need to do is throw in a few colourful tikz pictures. Progress of a sort:-)

@DavidCarlisle Heh heh heh... I actually did impress him with my mathematics in my analysis of colour compound distribution (ratios of red yellow versus yellow red, green blue versus blue green, etc.), but seeing as how he's a linguist and not a mathematician perhaps that is not such a feat. :p
@DavidCarlisle (it was a great feat for me in any case!)

@Plergux Sumir eru málfræðingar og stærðfræðingar

ooh partial derivatives

@DavidCarlisle Sumir já. Það er fámennt en góðmennt í þeim hópi. :p

@Plergux Eintómt líf sem ég lifi

2:54 PM
@DavidCarlisle Innantómt? Einmanalegt? "Eintómt" er "nothing but", e.g. "Það eru eintómir rugludallar á tex-latex-and-friends." :p

baa

@Plergux @PauloCereda struggles to understand my Portuguese sometimes as well. As I use an infallible linguistic oracle I can not imagine why it is hard to understand.

@DavidCarlisle oh

@PauloCereda small test: Panquecas de pato chinesas

@DavidCarlisle Pshaw, we've got "linguistic licence" :p
@DavidCarlisle Eu não entendo o que ele está confundindo! :p

3:10 PM
@DavidCarlisle OH NO

2 hours later…
5:40 PM
DEUTSCHE POST

6:03 PM
Nov 20 at 19:39, by David Carlisle
@LaTeXereXeTaL you are making the fundamental error of assuming that anyone has any idea what @PauloCereda is talking about most of the time.

6:52 PM
@PhelypeOleinik Expect a PR pretty soon: I thinkl I've sorted the xparse final move for 2e

@JosephWright Ah, cool!

@PhelypeOleinik Oddly, I'm a bit nervous about rollback

@JosephWright No idea why ;-)

@PhelypeOleinik I plan to leave the latex3 side until after the kernel release: it's tidying-up and should be safe left until then

@JosephWright Sounds reasonable, yes
@JosephWright Did you manage to move the commit history along?

6:56 PM
@JosephWright On success: Another great triumph for Great Britain! / On failure: Drat, Joseph broke it again.

@PhelypeOleinik The xparse one? No: it's really tricky to do that as it's a partial file, not even a full directory. We have the xparse history in latex3, and it's not going anywhere

@JosephWright No problem. I remember trying to move a commit history once and though I was too stupid to get it right :)
@JosephWright Shouldn't it go into develop?

@PhelypeOleinik @UlrikeFischer's sword kept the histories in ho-tex :-)

@PhelypeOleinik So it should: fixed
@PhelypeOleinik I hope I've got roll forward right, but we likely need to think about where best to 'park' the frozen xparse.ltx

@DavidCarlisle A history of blood and terror complaints and broken code

7:04 PM
@PhelypeOleinik blame @UlrikeFischer

@DavidCarlisle I'm not saying it's not doable
@DavidCarlisle, @PhelypeOleinik The issue with xparse is we actually don't want the full file anyway. So it's a partial re-write and move, at which point I'm not sure if doing a copy-commit-split approach is helfpul

@JosephWright it doesn't really matter much if the history is still in the old place. I was hoping to dwindle oberdiek to nothing so needed to move the history really (and it was whole files of course)

@DavidCarlisle True

If anyone was into jazz, some friends hold a live e-concert just now from one of the coolest jazz clubs in Europe :) youtube.com/watch?v=_gPrAEfzyzw

@DavidCarlisle We do have the history back though the old move from l3in2e of course

7:08 PM
@yo' ooh thanks

@JosephWright the secret fact about source control is the history is hardly ever used:-)

@DavidCarlisle Reminds me we should look at my color PR

@PauloCereda I know the drummer and the soundmaster

@yo' cool!

@PauloCereda ^^ at a concert!

7:18 PM
@yo' ooh

@PauloCereda lolwhut?

@Plergux no idea
@Plergux but it's Snake 'n' Bacon! :)

@PauloCereda Well, I've never had snake. For all I know it's just as delicious as bacon. :p

1 hour ago, by David Carlisle
Nov 20 at 19:39, by David Carlisle
@LaTeXereXeTaL you are making the fundamental error of assuming that anyone has any idea what @PauloCereda is talking about most of the time.

@Plergux LOL
@DavidCarlisle oh

7:35 PM
@PauloCereda bacon and dangernoodles sounds delish. :p

@Plergux ooh <3

@DavidCarlisle Well, I haven't got the foggiest what I'm talking about half the time either so that would break even, I think. :p

@Plergux <3

Came into the room just in time to see my name. Hope I'm not in trouble.

@LaTeXereXeTaL :)

2 hours later…
9:21 PM
@DavidCarlisle Hi, as I know you belong to most things related to TeX, just for fun (and to try to understand how it works), try to compile the examples of bidi-atbegshi and they don't work :( ...the team of "persian" TeX is related to the team of LaTeX3?
@PauloCereda I miss my swap partition :D

@PabloGonzálezL Some of us got 'asked' to help out
@PabloGonzálezL bidi is 'interesting'

@JosephWright Yes, quite interesting, I have never written anything LR, but the way the code works is what I would like to try to understand, especially why only XeTeX and not LuaTeX
@JosephWright I assume that bidi-atbegshi is another victim of hooks :)

@PabloGonzálezL It's not just interesting for that reason: I meant specifically code released as 'bidi', although yes, RTL is 'fun' [even more so for vertical setting: what does 'hbox' even mean in that context ;)]

@PabloGonzálezL what do you mean? which examples?

@UlrikeFischer From git repo

9:31 PM
@PabloGonzálezL yes, but which one? this bidi stuff is wandering all the time some like sand dune.

@PabloGonzálezL this gh organisation is me more or less :-) github.com/bidi-tex but it no longer looks after bidi and xepersion (a long story you don't want to know)

@DavidCarlisle, @PhelypeOleinik, @egreg, @MarcelKrüger, @UlrikeFischer (@PauloCereda) I wonder if Bruno's seen alanshawn.com/tech/2020/08/03/l3fp-accuracy.html

@JosephWright Wow!

@UlrikeFischer All test fail :(

@egreg Bruno did an amazing job

9:34 PM
@JosephWright Indeed!
@JosephWright Now, if he could improve the trig functions… 😉

@egreg Hyperbolics?

@JosephWright The blog mentions the trig functions being the less accurate of the lot.

@JosephWright hmm he's comparing against the double not the (newer) ieee standard decimal, still results pretty good:-)

@PabloGonzálezL yes that quite probably won't work with a current latex. bidi patches lots of packages and kernel code but seems no longer to update its patches. If we hadn't added something to firstaid, it wouldn't work at all github.com/persiantex/bidi/issues/18

@egreg True, relatively of course
@DavidCarlisle Well yes: the decimal standard was the aim; I just thought it was a nice article (I was re-reading his expl3` tutorial)

9:40 PM
@JosephWright Bruno was targetting the decimal floating point from IEEE-754-2008 not double, it would be interesting to check what it says about the bounds on trig accuracy.

@UlrikeFischer OK, I don't really understand the organization of this in Git (github.com/bidi-tex) or (github.com/persiantex)

@PabloGonzálezL bidi-tex is basically archived. At some time bidi was there but as I wrote it wandered around and you don't want to know the story (and we don't want to tell it).

@UlrikeFischer Heh-heh..OK
@DavidCarlisle To all this, some advertising for 'latex' in (texfaq.org/FAQ-tutbitslatex) would not be bad :)
I refer to 'www.learnlatex.org'.

@PabloGonzálezL ooh JUST READ THIS: IGNORE ALL FALSE PROPHETS TUTORIALS
4

@DavidCarlisle I will add this sentence to my list :)

9:57 PM
@PabloGonzálezL @JosephWright I guess I should add a line.

@DavidCarlisle at the begin and in bold ...

@DavidCarlisle And if possible repair the broken link (tac.dk/cgi-bin/info2www?(latex)) :P

@PabloGonzálezL looks like it's gone? tac.dk/?s=latex

@Dr.ManuelKuehner Hi, very kind. I hope you are well as all your loved ones. I send you and everyone a big greeting and a virtual hug. Sebastiano
Good evening from East Sicily to everyone.

@DavidCarlisle We probably should revise that page; it's quite out-of-date

10:05 PM
@DavidCarlisle Searching for broken links is a headache, there should be some script that does this efficiently, I add in Perl or Lua, no java of course (@PauloCereda)

@JosephWright yes I removed the para with the broken link and added something for learnlatex but a rewrite wouldn't hurt (eg listing some of the ones you found while starting the learnlatex effort)

@DavidCarlisle Probably best if someone else does that; might be rather biased otherwise (I wonder about @barbarabeeton)

@JosephWright :-)

10:53 PM
@Plergux -- Well, you could try to impress him with structure trees and parsed utterances.

11:19 PM
@JosephWright -- You rang? What would you like me to read?