TeX, LaTeX and Friends

6:38 AM

@JosephWright would you say that parsing class and package options with \keyval_parse:NNn is ok? (it would lose the information whether the argument was braced or not, so the "unused global options" error could be slightly changed) Also, is it correct that the class option list can't contain braces?

Joseph Wright

7:27 AM

@Skillmonlikestopanswers.xyz That's what l3keys2e does ...

Skillmon likes topanswers.xyz

8:08 AM

@JosephWright l3keys2e doesn't parse with \keyval_parse:NNn but with a \clist_map, iirc (well, but you wrote it, so your memory might be better in this case)

@JosephWright also, l3keys2e assumes standard category 12 commas (which doesn't seem like a bad thing at this place, active commas during class or package load time would most likely break a few things...)

David Carlisle

@Skillmonlikestopanswers.xyz the real problem is that the standard option lists go through \zap@space which does.. stuff we did trial a version that didn't do that (there's a gh issue) but it broke things and so we backed it out. probably at the least \documentclass and \usepackage should save the "raw" arguments as supplied in the document so an alternative option handler can use them, but as always there are compatibility issues even with that.

Skillmon likes topanswers.xyz

@DavidCarlisle which compatibility issue could arise if LaTeX did provide \raw@opt@file.ext as a raw version in addition to the zapped \opt@file.ext? (well I could think of a non-optimally build \@unusedoptions list which would break in conjunction with older packages if the class author screwed up).

@DavidCarlisle is the \zap@space code also the reason why \documentclass[foo={my option}]{article} breaks?

David Carlisle

8:23 AM

@Skillmonlikestopanswers.xyz I forget now, something broke:-) something that was patching the \documentclass code failed to patch just because we'd made (any) change I think

@Skillmonlikestopanswers.xyz that doesn't necessarily break (I think) but zap@space iterates through the list in an edef with a space delimited argument so really any commands in the val part of a keyval list are suspect and braces can be dropped at weird places depending on where otherwise ignores white space appears, = {zzz} for example would leave {zzz} as the argument of a space delimited macro so the braces would be dropped,

Joseph Wright

8:45 AM

@Skillmonlikestopanswers.xyz Well yes, but that's much the same idea really

Skillmon likes topanswers.xyz

@DavidCarlisle I tested this morning and both article and scrartcl break for a class option with braces as the value (but I don't recall the exact option list I tested with).

Ulrike Fischer

@DavidCarlisle I don't think that something really broke, but nobody had enough time for real tests (e.g. with classes like KOMA which do additional option handling).

David Carlisle

@Skillmonlikestopanswers.xyz yes but that's probably \in@ (look what it does:-) but that's not so much of an issue as a class doing its own option processing can avoid that. but currently the main option list handling is already broken for keyval parsing.

@UlrikeFischer possibly, yes

Skillmon likes topanswers.xyz

@DavidCarlisle I blame you for it :)

@DavidCarlisle My test package doesn't break for braced arguments in the option list...

@DavidCarlisle and neither does my test class (which doesn't touch option parsing at all but just does a \LoadClass{article})

Ulrike Fischer

@Skillmonlikestopanswers.xyz personally I don't like long option lists in the documentclass. When there is only \documentclass{scrartcl} I can double click on the documentclass and it opens the source, when there is \documentclass[millions-of-key=value-on-various-lines]{scrartcl} it fails. I normally put at most the language, draft and the fontsize there and for everything else use \KOMAoptions.

Skillmon likes topanswers.xyz

9:03 AM

@UlrikeFischer I personally use very few options as well, but currently I'm trying to plan out a class/package option parser based on expkv (because that's what's still missing compared to the main competitors) so I can't just assume my own behaviour.

Paulo Cereda

9:20 AM

quack

Paulo Cereda

9:38 AM

'Doctor Who' Cosplay Contest w/ Jodie Whittaker & David Tennant

►

@Skillmonlikestopanswers.xyz ooh expkv-opt

David Purton

1:10 PM

Hello!

I just got around to noticing this: github.com/dcpurton/pdfoverlay/issues/1 asking me to register in l3prefixes.csv

Is 'Last Updated' meant to be when the package was last updated? i.e., in my case: First registered will be 2020-06-22 and Last updated will be 2019-11-03

Or should Last Updated also be 2020-06-22?

Joseph Wright

@DavidPurton Last updated would be a change in e.g. Git repo or issues location

David Purton

Ta, so both same day to start with.

Joseph Wright

@DavidPurton Exactly

Skillmon likes topanswers.xyz

1:43 PM

@PauloCereda Andrew was best :(

yo'

2:00 PM

Hi, please, what is the preferred way of testing the engine and throwing an error in an expl3 class if the engine is wrong?

Ulrike Fischer

@yo' \sys_if_engine_XXX:TF and some error?

yo'

@UlrikeFischer ah right, thanks!

daleif

2:14 PM

In biblatex with no additional options, if we add urldate = {2020/06/22} to say a @book entrty, why is the urldate silently ignored (I see '(Visited on )` in the output. There are no mentions of this being an issue in the biber log

Here is a full MWE I made for a question on the site. I get the above via TL2020

\documentclass[a4paper]{book}

\begin{filecontents}[overwrite]{\jobname-refs.bib}
@incollection{Slemrod.2002,
author = {Slemrod, Joel and Yitzhaki, Shlomo},
title = {Tax Avoidance, Evasion, and Administration},
urldate = {2020/06/22},
pages = {1423--1470},
volume = {3},
publisher = {{Elsevier/North Holland}},
isbn = {9780444823144},
series = {Handbooks in economics},
editor = {Auerbach, Alan J. and Feldstein, M.},
booktitle = {Handbook of public economics},
year = {2002},
address = {Amsterdam}
}

Ulrike Fischer

@daleif wrong format? urldate=2020-06-22

daleif

@UlrikeFischer probably, but when why doesn't biber complain in the log? It complained with the OPs original 22.06.202

22.06.2020

Ulrike Fischer

@daleif good question. I get a warning when I use urldate={blub} but not with the 2020/06/20. But it is the source of the problem.

daleif

Blame who ever is maintaining biber?

Was the xx/xx/xx syntax supported at one point?

Ulrike Fischer

@daleif no idea. I would open an issue and ask github.com/plk/biber/issues

daleif

2:23 PM

@UlrikeFischer seems / has a very special meaning in biblatex dates

Ulrike Fischer

@daleif right it is about ranges. But then why does it disappear?

daleif

@UlrikeFischer exactly, it should at least warn

On the other hand it might be the template that does not check the given data correctly

I'd probably need to rewrite the wording on my biber issue.

Ulrike Fischer

@daleif no the urldate is really not there in the bbl. but with 2020/2021 is works. I think this is a bug in the date parsing.

@daleif the same happens with date date = {2002/06/01},: no warning but no date either (and an empty year).

Paulo Cereda

4:42 PM

Guys, do you know that if you have an alias, say, alias ls=colorls and want to use the original ls, you simply go with \ls?

David Carlisle

@PauloCereda yes

Paulo Cereda

@DavidCarlisle oh I didn't know about it :(

David Carlisle

@PauloCereda you are only a duck

Paulo Cereda

@DavidCarlisle oh

/sad quack

David Carlisle

@PauloCereda don't worry, soon be dinner time

Paulo Cereda

4:46 PM

@DavidCarlisle phew

Oh wait

@DavidCarlisle oh no

Stefan Kottwitz

5:37 PM

@PauloCereda aliases? do you have alias lolwut=sudo? Then if some command fails, typing lolwut !! helps, since !! expands to the last used command. Want my full aliases list?

Helps to annoy people who are looking over your shoulder. (boss)

Stefan Kottwitz

5:53 PM

and yes, I escape it by prefixing with \ to restore the original meaning.

root@latex.org:~/# ls -al /usr/bin/lolwut
lrwxr-xr-x 1 root root 6 Oct 3 2020 /usr/bin/lolwut -> pdftex

Skillmon likes topanswers.xyz

8:33 PM

@PhelypeOleinik cool new package and a great idea!

barbara beeton

8:51 PM

@DavidCarlisle -- There's a problem reported at tex.stackexchange.com/q/550638 that may be left over from the relatively recent move of the TUG box to a new IP address. (It hit me; the previous owner of the IP address had been a spammer, and the IP address had been banned. After Karl explained to the ISP, the address was cleared.) Please add comment, tell OP to send a message to office@tug.org.

user1271772

Has anyone here ever seen multirow{1}{*} ? I'm asking here because I'm not sure if it's enough to ask as a question on the main site. I have googled "multirow{1}{*}" and not found any results.

David Carlisle

@barbarabeeton linked to your comment

@user1271772 multirow{1} seems rather pointless but I guess it's legal

Phelype Oleinik

@Skillmonlikestopanswers.xyz Thanks :-)

barbara beeton

@DavidCarlisle -- Many thanks. This is a great nuisance!

Skillmon likes topanswers.xyz

9:23 PM

@PhelypeOleinik next logical step: Building it into l3docstrip so that it doesn't have a runtime impact :)

@PhelypeOleinik regarding things like \prg_new_conditional:Npnn, how about an environment like structure which can be used around a single definition. That way you don't have to parse wether there are any more arguments which are grabbed, but you can grab the environment body and do the replacement there (with the first occurrence of every #[<name>] stepping the used number).

Phelype Oleinik

@Skillmonlikestopanswers.xyz Hm... It would be nice. Though the (executable) code would be unreadable.

@Skillmonlikestopanswers.xyz It's an idea... Though I'm not keen on adding more commands to the package. I might write some extensions loadable through package options (like the \newcommand interface in Denis' issue)

@Skillmonlikestopanswers.xyz But thanks! I'll think about it

@Skillmonlikestopanswers.xyz (the answer is probably no, but) do you happen to know if there's a trick to detect a \global prefix after it has been consumed by TeX?

@Skillmonlikestopanswers.xyz \global doesn't leave any clue in the macro, as \long does, so unless a group is started for the sole purpose of detecting \global, I couldn't find a way

David Carlisle

9:41 PM

@PhelypeOleinik I don't think there is anything reasonable you can do (short of implementing tex-in-tex but that's ~~crazy~~ Bruno

Phelype Oleinik

@DavidCarlisle But even doing that, catching \global after TeX absorbed it doesn't seem to be possible

David Carlisle

@PhelypeOleinik but if you were using something like unravel from the start not just after \global` may have been used you could presumably catch it, but you'd just be emulating tex execution in tex not actually executing it so not that useful

Phelype Oleinik

@DavidCarlisle Yes, that's what I thought. My idea was to make something like \global\named\def... work

@DavidCarlisle Nothing that adding a new primitive doesn't solve ;-)

David Carlisle

@PhelypeOleinik \edef\global{\something\originalglobal} ?

Phelype Oleinik

@DavidCarlisle People wouldn't be too happy if I redefined \global

Ulrike Fischer

9:51 PM

@DavidCarlisle which reminds me that Hans commented some explanations about the new argument handlers in ltmx with "but then of course also less possibility to 'show off how clever one is wrt writing obscure macros' "

David Carlisle

@PhelypeOleinik if you asked most users they'd probably say they would not be happy if latex redefined \par or \input, the trick is not to tell them.

4

Phelype Oleinik

@DavidCarlisle :-)

@DavidCarlisle Hm... \afterassignment seems to hold just one token, different of \aftergroup. If it stored more, that would actually work great...

Henri Menke

10:07 PM

@JosephWright We're experiencing CI errors in PGF.

(/tmp/texlive/texmf-dist/tex/latex/l3backend/l3backend-dvisvgm.def
! Illegal parameter number in definition of \__draw_backend_color_fill_rgb:n.
<to be read again>
2
l.391           { \fp_eval:n { 100 * (#2
                                      ) } }

Ulrike Fischer

@HenriMenke ups.

user1271772

10:27 PM

@DavidCarlisle what about the {*} in the multirow{1}{*} ?

Transcript for

$TeX - LaTeX$ TeX, LaTeX and Friends