@PauloCereda I'm sorry, Paulo. I had to go and eat. I'm going to restart and try arara. It's my fault. I just didn't know I was running a version-specific java. (Didn't know this was possible without AUR until @Skillmon explained it.)

@PauloCereda It works! <3 @Skillmon @samcarter_preparing_for_xmas Thank-you!

@PauloCereda Actually, I'm not sorry arara broke. (1) I can manage without it (so better than, say, biber breaking.) (2) I suspect this will solve a mystery in LibreOffice which has been annoying me, but not enough to get me to do anything about it. :) [I can manage without arara, but the frustration level without it is very high!]

@cfr I am very happy to hear you are well and safe. :)

We hwyadens are very good at Java versions. :)

@cfr on Tuesday, we were talking about Flanders and Swann and Alan mentioned to Barbara that [Donald] Swann was Welsh. And then I mentioned that I know how to write Cymru and hwyaden. And then we thought of you and how much we missed you in the community. :)

Morning, all!

@yo' hi Tom!

@yo' breakfast

@DavidCarlisle oh... no?

@PauloCereda Dobré ráno

@DavidCarlisle ooh

@DavidCarlisle quack

@DavidCarlisle yes please.

@yo' duck?

@DavidCarlisle oh no

@DavidCarlisle as Alex reminded me yesterday atenddvi is no longer needed/used in a new format, so I made it a no-op, froze the code similar to hopatch and adjusted the documentation. I will upload it in the afternoon, OK?

@UlrikeFischer Good plan

@UlrikeFischer sounds good thanks

twitter.com/WolfBruening/status/1326821248320630785

@samcarter_preparing_for_xmas OMG

@samcarter_preparing_for_xmas I should try this :)

@samcarter_preparing_for_xmas mail. :)

@PauloCereda wow! We should all do that at the next TUG meeting :)

@samcarter_preparing_for_xmas YES

@PauloCereda (also got your other mail, will reply as soon as I have tested)

@samcarter_preparing_for_xmas Oh no worries, I will mark the rule as working untill a report is received. :)

@PauloCereda Ah, Schrödingers rule: works as long as someone tries it :D

@samcarter_preparing_for_xmas LOL

@samcarter_preparing_for_xmas rule updated, we decided to remove some daemon references.

@DavidCarlisle I prefer English breakfast.

@PauloCereda you got mail ...

@UlrikeFischer got it

@UlrikeFischer OH MY GOD

I want mail, too :(

Is there anything 007 can't infiltrate? ^^^

@Skillmon /poke

@Skillmon LOL

@UlrikeFischer hah!

@Skillmon same thoughts ;-)

@UlrikeFischer but I was 8 seconds faster :P

@Skillmon you were probably slower, as you typed more but started earlier ...

@UlrikeFischer But I typed mine in VIM, so most likely I was faster because VIM.

@AlexG it looks as if ghostscript is breaking more and more dvips/ps2pdf code .... tex.stackexchange.com/q/570598/2388

Can I fill colours inside a cell in LaTeX table? And Does Stack exchange support it?

@YouKnowMe Do you mean like tex.stackexchange.com/questions/50349? Not sure what you mean by SE supporting it.

@YouKnowMe if you mean mathjax in a website then it doesn't support colortbl (we don't support mathjax at all on this stackexchange site) you can have coloured background in mathjax, see docs.mathjax.org/en/v2.7-latest/tex.html but not directly ties to an array cell. (and mathjax only does math so only array not tabular anyway)

@UlrikeFischer In fact, -dNOSAFER is not a new requirement for file access with gs. Though it can be sidestepped, by anticipating file embedding to the DVI-->PS step. You may remember perhaps, we discussed this here a while ago, pdfbase implements this for creating file stream objects.

@AlexG yes I remember (and know) about -dNOSAFER. But forgot it in the first test. Now I only wonder if the OP forgot it too, or if there is nevertheless a problem.

@UlrikeFischer I'll try later this evening.

@AlexG I would wait for a reaction of the OP.

@UlrikeFischer Ok, will do that. Nevertheless, it's time for an update of GS here on my home laptop. (A test with gs-9.52 worked fine.)

@AlexG don't you have it in texlive?

@UlrikeFischer No, TL for Linux ships without GS. Does it?

@AlexG really? I have it in C:\texlive\2020\tlpkg\tlgs\bin, and it is 9.53.3. It is a constant source for problems as it somehow finds resources from the external ghostscript and fails if the versions differ. So everytime texlive updates I have to update the external ghostscript too.

@JosephWright Can you add the this question to the close message of this question: tex.stackexchange.com/q/570519/2693. For the specific problem it's a much better duplicate, since the main duplicate is more generic.

@StefanKottwitz Hi Stefan, since you're here before @JosephWright see ^^^

@AlanMunn I'll check it! No hurry, since it's of 2011 :-)

@StefanKottwitz No, it's not. It's yesterday.

Oh the other one is new, right!

Mac people, have fun with Big Sur

@StefanKottwitz The idea is to add the question I linked to in the comment (and mentioned in the comment by jsbibra) to the list of duplicates in the closed question.

@AlanMunn yes, it's in the first comment, I'll check how to add it in the best way

@PauloCereda We Surely will. Or maybe it will make us Surly.

since messages are auto-generated

@AlanMunn :)

@PauloCereda when I google Big Sour it only shows me Mac stuff (yes sour)

@StefanKottwitz That's good :)

@PauloCereda So nobody knows The Big Sour? The it will be my next book or movie or trademark. Big thing after the Big Sur is forgotten after the next Mac release.

and anybody googling Big Sur then, will arrive on my site >:-)

@UlrikeFischer There is no tlgs directory. Hence no gs as part of TL, let alone ps2pdf and friends. (I have a full installation.) Even dvisvgm dynamically links the system's libgs.so.

@AlexG yes, confirmed.

@AlexG so it is windows specific.

@UlrikeFischer Also, TL on Windows ships with a TeX/LaTeX editor by default, as far as I remember.

(TeXWorks?)

@AlexG yes, correct. :)

@PauloCereda Thanks <3

@AlexG quack <3

@Plergux nice ;-)

@UlrikeFischer (^-^)

@PauloCereda I'm sorry. Really. I'm bad at this ....

I can't even remember how to use forest.

[Though that's a failure of a different kind.]

@cfr quack!

@Plergux ooh a secret

Hey, is \immediate\write18 broken in LuaLaTeX? Or are there cases when it doesn't work? I just checked @JosephWright's answer at tex.stackexchange.com/a/20446/11002 but it doesn't indicate there should be an issue

@yo' not broken it just writes to the file on channel 18

@yo' @JosephWright's answer is from 2011 or pre-history in luatex terms. add the utterly brilliant shellesc package and you will get \write18 back as well as the more sensible \ShellEscape{..} syntax.

@DavidCarlisle :) Ok thought so. But thanks for confirming.

(The thing is, that answer is the first google hit for "lualatex immediate write18", so it's probably worth updating...)

@yo' when they increased the number of output streams from 16 to 256 they dropped the special nature of 18 as it is rather odd/dangerous to have one special stream in the middle of the run

@DavidCarlisle I understand the logic, but I feel old in these things: it used to work, reliably, on all engines :(

@yo' I remember the time before the feature was added:-)

@DavidCarlisle you mean before \write18 itself?

@yo' lualatex always skips 18 in luatex when you do \newwrite so that it is available, and if you add shellesc package then it redefines \write to check for 18 (and use \directlua{os.exec...}

@yo' yes

@DavidCarlisle ah thanks, I can feel young again :-)

@yo' but in plain luatex \newwrite will just allocate 18 in the usual sequence

@DavidCarlisle yeah, I was wondering how do you do it, and it's cool that 18 is skipped in lualatex

@yo' excellent code, it is:-)

@DavidCarlisle why is it \write18 and not \write16?

@UlrikeFischer I believe 16 is log, 17 is terminal+log

@yo' everything outside 0-15 is terminal, try \write9999{hello}.

@UlrikeFischer because \write to an unallocated or out of range stream types to the terminal (that is how \typeout works) so people already had \immediate\write16{hello world} and it wasn't safe to suddenly make that execute code, but a search of existing corpus (pre google) suggested it was safe to pick 18 as it was not use anywhere

@DavidCarlisle and leaving out 17 as safety distance?

@UlrikeFischer does "one for luck" translate? :-)

@DavidCarlisle :-)

and btw, just to make sure I remember correctly: conditionals inside tabular are a highway to hell, right?

@yo' if they go over more than one cell

@yo' better to think of them as an interesting twisty mountain path which is fun but there is a chance you fall into a chasm

@UlrikeFischer yep, they do :) (or rather they would)

@UlrikeFischer should I blame Nicola or you?

@DavidCarlisle wouldn't \addtocounter{table}{-1} be safer?

@DavidCarlisle you can blame me not to have seen the question. When I see "table has wrong number" I always look for a longtable ;-)

@yo' well i wondered about that but you might have two tables-of-contentsy things and you really want the first table to be 1 not 1 less than whatever wrong number you got to

@DavidCarlisle yeah, that's right. Anyway, if you need long glossaries, your tables should be numbered within chapters anyway -- problem solved, you can blame the user!

@UlrikeFischer the initial version of the question just showed a table no mention of longtabl or glossaries or anything. then people on meta wonder why we ask for a MWE :-)

@DavidCarlisle yes, but it is a simple rule: if the table counter is doing odd things there is a longtable around.

@UlrikeFischer the very first version of longtable used table counter to generate the internal csnames for each table, but I changed that as soon as I tried a class that reset the table counter per chapter and added a private internal counter, so from then on longtable didn't need to increment table (and \caption could do it as normal) :

% \changes{v3.02}{1992/04/09}
%    {(Michel Goossens) Longtable fails if the table counter is reset
%     during a document. Now use an internal counter, but still
%     increment table so \cs{caption} and \cs{label} work out right.}

Oh we can blame Michel for his feature request!

@UlrikeFischer "wrong behaviour" is not something I associate with my code, as a rule

@DavidCarlisle Well, you could also check for the presence of \caption :-)

@DavidCarlisle it is naturally user error if longtable is not handled correctly.

@yo' because doing conditionals inside a table is easy?

\def\longtable{%
\let\LT@origcaption\caption
\def\caption{\gdef\LT@hascaption#1{#1}\LT@origcaption}%
\gdef\LT@hascaption#1{}%
\stepcounter{table}
...
}

\def\endlongtable{
...
\LT@hascaption{\addtocounter{table}{-1}}%
}

@DavidCarlisle or you could have rewritten all the tabular code without \halign in 1992; there's a good chance it would have caught up in the last 28 years :-)

@yo' no you need it set earlier so \caption{\label{..} gets the right value, and worry about\caption appearing in both the first head and the main head, or perhaps in the table body.

@DavidCarlisle yeah, realized that; tried to correct, was too late. Now the true and false branches are switched.

@yo' in v5 I was playing with an option so you could say captype=none to disable the counter or captype=figure if you were using longtable for alignment but want it to be figures etc

@yo' You should use \refstepcounter, and don't forget hyperref, which doesn't really like counters that go back by one.

@UlrikeFischer yeah I know, excuses excuses. My code is perfect and vote Donald Trump