@TorbjørnT. I can't stop laughing!

@VincentMiaEdieVerheyen no you want \percentifier{#1} not ##1 but if you want help with an error message you should always provide an example that generates the error. As I said here a day or so ago it will work if you put it in the right place so just positing working code and saying that you get an error if you change it isn't the best way to get help.

There's an arara update to CTAN, but don't get your hopes up! It's just a maintenance release because of the Java 9.0 specification. :)

@PauloCereda have you finished arara 4 your thesis?

@DavidCarlisle oh no

@PauloCereda I scroll and scroll, and suddenly, on page 29 I can read about arara. Seems to be a nice tool!

@mickep awww thank you! <3

@mickep Except you are using TeXworks and are used to kill a typesetting process using the Typeset button...

@TeXnician oh no :)

@PauloCereda That seems to be the best description ;)

@TeXnician I think Windows is the one to blame here, at least to some extent. :)

@PauloCereda Interestingly enough, because I have issues on Win10, Win8, Linux Mint and Arch. All of them won't kill the arara process (including the spawned latex).

@TeXnician Ah, in that case I will investigate. I think I know where to start! Thanks for the pointer!

@PauloCereda You're welcome. I thought it was a design flaw in terms of TeXworks' killing...

@TeXnician I don't really know who to blame on this, but I can investigate. :)

@TeXnician: can you easily reproduce the issue?

@PauloCereda Well, with some big projects and some aux file corrupting failure yes, but I'll try to boil it down.

@TeXnician Oh please, if you could come up with a MWE, it would make my life so much easier! But no worries, take your time!

@TeXnician I happen to still be in love with my terminal, so I compile from it...

And my kids love when the output in the terminal is flickering by, and then all of a sudden the pdf updates with new figures :)

@mickep Oh, I also love my terminal, when I'm on a reasonable platform (Linux). But I hate it, when on Windows...

@TeXnician Oh, I've never used TeX on Windows (and it does not sound like I should).

% arara: lualatex
% arara: lualatex: {synctex: 1}
%!TeX TS-program=arara
\documentclass{article}

\begin{document}
	Test\bla
\end{document}

@PauloCereda ^ Compile and abort using Typeset button when the error prompt shows up.

Tested with TeXworks 0.6.2 (AUR) on Manjaro Linux.

@TeXnician Thank you! <3

@TeXnician: give me a minute.

@PauloCereda Verified using Win10 (and up-to-date TeXlive TeXworks).

@TeXnician: reproducible! Thank you!

@PauloCereda I guess you tested on Fedora?

@TeXnician Yes.

@PauloCereda And who is to blame?

@TeXnician Let me try just one more thing and I will reveal. :)

@TeXnician just use a reasonable shell, I usually use bash running in an xterm on windows and it's not so different to being on linux

@DavidCarlisle Unfortunately I'm too lazy to care about my Windows installation (except for updates). Usually booting up Linux is the better idea :)

@TeXnician I can give you a clue who to blame, begins with P ends with o and is an anagram of oluaP

@DavidCarlisle Let's wait for him realizing that too ;)

@TeXnician I have no problems with terminals on windows

@UlrikeFischer I regularly have to SSH onto Linux servers and perform some actions there. I have had many issues with that within a Windows CMD in the last few years.

@UlrikeFischer Neither do I. The explanation is simple enough: I don't use windows.

@TeXnician ^^^^ ;-)

@UlrikeFischer You're definitely a Linux user masked as a Windows user :D

@UlrikeFischer @TeXnician note how @UlrikeFischer hasn't actually managed to process any document in any of those command windows

@DavidCarlisle :)

@DavidCarlisle I managed to hide my data before making the screenshot ;-).

@UlrikeFischer Undefined control sequence? ;)

When I try to upload an image with an answer I get an error "Failed to upload image; imgur is rejecting the request"

is this a problem for other people as well?

@PauloCereda Did you find some duck to blame?

@TeXnician I did. :)

@TeXnician: It's tricky because TeXworks really hangs the arara process. However, the hooks are in place and, if given the correct signal, arara destroys the associated processes. If executed in the terminal and terminated with a SIGINT, arara closes everything correctly. I must investigate how TeXworks deals with this.

for tex.stackexchange.com/a/395328/89417 I was wondering if I should put the defs in the preamble or not

any thoughts/best practices?

@DavidCarlisle See Chris is on form: meeting will be fun!

@JosephWright oh

@JosephWright: wait for @DavidCarlisle to see my terminal

@PauloCereda :)

@Marijn I would say yes, they should go into the preamble. Stuff after the \begin{document} is supposed to be the content not the markup definitions.

@AlanMunn Agreed though gets tricky with non-English input in pdfTeX as inputenc only kicks in AtBeginDocument

@AlanMunn that would be the general principle, but in this case the definitions are actually content and not markup (being text-only macros)

so I reconsidered

@JosephWright No. inputenc makes them directly active. It's babel that kicks in only at begin document.

@UlrikeFischer Oh yes, silly me

@Marijn Sure, but they're still markup within the document even if they are constants.

@UlrikeFischer Same general point though: you can't be sure you'll get the 'right' outcomes (oh for a new format ...)

@JosephWright I hadn't thought of that, as I don't use pdftex anymore for my documents.

@JosephWright Well you never can really be sure to get the right outcome. Did you ever tried this:

\documentclass{article}
\usepackage[utf8]{inputenc}
\usepackage[thai,ngerman]{babel}

\begin{document}
Grüße
\end{document}

@UlrikeFischer Strangely no ;)

@JosephWright ;-). Well the thai.ldf ends with lots of lines like \catcode`\^^a1=11. You can imagine the side effects.

@UlrikeFischer Oops

@UlrikeFischer Cute. :)

@JosephWright Exactly my reaction ;-) (I made a bug report but never got an answer).

@UlrikeFischer :(

@UlrikeFischer The team should probably look to fix that I guess (doable if there is no response from the maintainers).

@DavidCarlisle ^^^

@JosephWright at the very least babel could check that it hasn't messed up the catcode regime even if we don't change the externally maintained def files

@DavidCarlisle Ah, good idea

@DavidCarlisle One for the team list

Volume all up.

@DavidCarlisle We could probably just loop over everything: on a modern system that's quite doable in pdfTeX (which is where this is aimed at)

@JosephWright oh for @@@ it globally resets all lccode and uccode as well

@DavidCarlisle Yes, I saw that, hence saying we might want some more 'focussed' action

@DavidCarlisle github.com/tlwg/thailatex/tree/master/babel pull request time?

@UlrikeFischer I tried to imagine how many things that breaks, but soon lost the count.

@DavidCarlisle @JosephWright this would then probably break the thai output. One would have to check the fontencoding. But someone wants to do something here: I actually even wrote an lthenc.dfu at the time of the bug report (in 2013) and could pass it on.

@JosephWright worth a try, if anyone is there...

@JosephWright since the uc/lccode table probably needs to break 2e rules to work with existing fonts probably least-effort change would be to keep the definitions but make them be attached to \thaitext rather than globally active, still wouldn't be fully conforming but at least @UlrikeFischer would be able to give greetings

@DavidCarlisle Sounds reasonable

so, why is the title of a question a link for itself?

(also on other SEs)

@Marijn I don't know why, but I sometimes use it to reload questions.

like F5? or 'Go' in the address bar on a phone?

@Marijn I use it to get the link to the question.

like 'share' below the question?

or copy in the address bar?

@Marijn well no that tags a user/id on the end and doesn't actually change the URL in the browser

@Marijn and that doesn't change the URL to the question either (if for example you reached the page from an answer link)

@Marijn but the real answer is that the site mechanics are whatever is implemented by the stackexchange company, and users and moderators on individual sites get little say. It is what it is:-)

:)

I know, I was just curious as to why

QUACK

Today is Annoy the CTAN guys day.

@PauloCereda as in "oops can you re-install that zip file that you just spent the last hour installing as I forgot something" ?

@DavidCarlisle worse: "oops I didn't read any manuals and simply submitted three versions of a complex package that got rejected + a lecture on each failed attempt on how to constructively fix it".

@DavidCarlisle I should include this link in the next communications

@PauloCereda ctanify is your friend here. Did you use it?

@AlanMunn Oh no, I forgot about it!

@PauloCereda Or have you joined the cool kids (@JosephWright) and are using l3build? (Actually I have no idea if it offers similar functionality, having taken @DavidCarlisle 's advice and not read the documentation.)

@PauloCereda Ever since I started using ctanify I have been lecture-free. :)_

@AlanMunn I have a mental block regarding l3build due to some traumas regarding PR. :) (@JosephWright)

@AlanMunn ooh

@PauloCereda if l3build and ctanify don't work, you could consider using arara

@DavidCarlisle :)

@DavidCarlisle ooh

@PauloCereda But forget TeXworks ;)

@TeXnician :D

@TeXnician: arTeXmis will fix everything. :)

what is @egreg going to do for his next phone? bbc.co.uk/news/technology-41551546

@DavidCarlisle LOL

@DavidCarlisle But remember theverge.com/tldr/2017/10/3/16412372/…

@JosephWright @DavidCarlisle: Is there a way to force either fp_eval:n or \fpeval from xfp to accept numbers with decimal commas instead of dots? E.g. \fpeval{2,5 * 3,5}?

@ChristianHupfer No, or at least not with expansion: these are strictly 'international' format expressions

@ChristianHupfer regexp replace , to . Bruno must have had some reason for writing a regex engine...

@DavidCarlisle cause pain. :)

@JosephWright's talk about l3regex:

@ChristianHupfer See tex.stackexchange.com/questions/372356/…

@JosephWright I had expansion in mind ... :-( I use a lot of decimal numbers in fundamental constants or simple calculations, the numbers having , and want to typeset the result of \fpeval calculations, and perhaps feeding it to \SI{}{\someunit}

:D

@PauloCereda Jokes from the Crypt - day, is it? :-P

@ChristianHupfer you are mean

@DavidCarlisle ^^ Christian is mean

@UlrikeFischer struth that was answered by me!

@PauloCereda: Christian is mean ; regex ; Christian is not mean ;-)

@ChristianHupfer ooh

@ChristianHupfer so many regex you could use: s/is/is not/ or s/Christian/Paulo/ or ...

@UlrikeFischer Thanks. I did not yet search on TeX.SE ... like a newbie :D @DavidCarlisle: I had something similar in mind, but was too 'scared' changing \num

@DavidCarlisle: I think my fourth submission will be rejected because I didn't include a README...

@PauloCereda yes

@DavidCarlisle ooh more email

@PauloCereda If you had had a README you could have read it and that would have told you to have a README (or they prefer README.txt or .md these days)

@ChristianHupfer It would be very tricky I think to cover both input forms, keep things expandable and not have issues with invalid values

@DavidCarlisle ooh more README

@JosephWright There are no invalid values in Physics :-D They are normalized away ;-)

It's like having a spare safe key which is kept inside the safe

@ChristianHupfer 1.23,4 might be :)

@PauloCereda safest place for it

@JosephWright Typos like that should make LaTeX shout at you, so such errors shouldn't be ignored anyway.

@DavidCarlisle ooh

@JosephWright although it would be easy enough to write an expandable comma to point replacement and then allow \fpeval{\commasusedinthatbitofeuropethatdidntvoteforbrexit{123,5 + 3,4}}

@JosephWright The link to @UlrikeFischer's question is quite useful -- she mentioned the reasoning of 'sanitizing' in a comment: I just have to use , instead of . in German school

@DavidCarlisle Yes, true

@JosephWright we could agree on that name at next team phone conf, might be easier than agreeing on utf8 handling.

@DavidCarlisle Also true :)

@DavidCarlisle but utf8 handling is important too. I will have to process utf8-csv files and so I am quite interested in a solution ;-) (even if it means to use \usepackage[only-utf8-and-nothing-else]{inputenc}).

@UlrikeFischer If I get something written I'll forward you a copy.

@DavidCarlisle Thanks. Did you got my mail?

@DavidCarlisle: I got that email: another rejection. :)

@UlrikeFischer on thai? yes thanks ( thought I'd replied)

@PauloCereda Thank you for your postal card. Not rejected!

@ChristianHupfer @UlrikeFischer vvv

200.95
200.95
200,95

\documentclass{article}

\usepackage{xfp}

\def\ctod#1{\xctod#1,\relax{}}
\def\xctod#1,#2\relax#3{\ifx\\#2\\\expandafter\xxctod\else\expandafter\xctod\fi.#2\relax{#3#1}}
\def\xxctod#1\relax{\fpeval}

\def\dtoc#1{\expandafter\xdtoc\romannumeral`\^^@#1.\empty}
\def\xdtoc#1.#2{#1\ifx\empty#2\else,\expandafter\xdtoc\romannumeral`\^^@\fi#2}



\begin{document}

\fpeval{123.45 + 77.5}


\ctod{123,45 + 77,5}


\dtoc{\ctod{123,45 + 77,5}}

% expansion

\typeout{\fpeval{123.45 + 77.5}}

\typeout{\ctod{123,45 + 77,5}}

@JosephWright might quibble a bit about my expl3 expansion idiom but whatever.

@DavidCarlisle In seriousness, we could add this concept either as an option or as part of \fpeval

@StefanKottwitz <3

@JosephWright yes main issue would be two-argument functions do we haveatan2(x,y) I xfp must admit I haven't checked...

@DavidCarlisle Ah, that is true

@JosephWright but basically commas for decimal point are unusable in a commas separating arguments setting so a user option for simple arithmetic in common natural language contexts would still be useful I think.

@DavidCarlisle Could well be

@ChristianHupfer how do you naturally write max(1.2,1.5) if you are writing 1.2 as 1,2 (serious question for once)

@PauloCereda Did you ever turn this smiley to the right instead of to the left? >:-)

or @StefanKottwitz or @UlrikeFischer or any other people using commas:-) ^^^

@DavidCarlisle We could ask Frank ;)

@DavidCarlisle Seriously, this is the issue with allowing commas and full stops: you end up with a syntax nightmare

@DavidCarlisle It's OK in siunitx as the aims are very different

@DavidCarlisle The usual way is to use ; instead of , as separator between the two numbers, e.g. $P(1,2 ; 1,5)$ if a point in a 2D space is meant

@ChristianHupfer Was my guess :)

@DavidCarlisle well, first, you input 1{,}2; then you can write (1{,}2, 3{,}5), but it's still risky and a semicolon is preferred.

@yo' using tex syntax like {,} would lose the point here which is to use "natural language input" if you need tex-specifc input you may as well use .

@DavidCarlisle ah, then max(1,2, 1,5) should do, but it's still quite risky

@yo' Just say $a = \num{1,2}$ and $b=\num{1,5}$ then use \max{a,b}, that should be 'unique' ;-)

@DavidCarlisle Devastating news

@DavidCarlisle yes I wrote about it in my question "... I would accept the requirement to use spaces here, or to use another symbol e.g. ;, or to use some protection command like \dontsanitize{max(1.5,2)}, or max(1,5 \realcomma 2)." But for "normal" calculations your commands are good, I could use them inside \num and SI without problems.

Wow! stackoverflow.com/help/badges/7358/…

@StefanKottwitz ooh

@DavidCarlisle: fifth attempt, here we go. :)