@user0 while I understand your point, deleting everything seems like way too much effort.

@Skillmon Right. That's why I put a premise "if anything else happens at all with the topic..." If you do anything else, it would create some bias.

@user0 and hiding should create no big bias. If anybody is interested he can easily see the cause.

Apparently a user, that has been suspended many times ... stackoverflow.com/users/285587/your-common-sense

@Skillmon Well, the mods would hide something of their choice rather than the whole discussion. It would introduce some bias.

@user0 but the point about hiding it is that users can still see it if they are really interested.

Citing: "I've been expelled from Stack Overflow solely because of my pursuit of quality. Beside many other things (like title and tag editing, finding dupes, etc), I tirelessly chased people who wrote bad answers, and naturally those people bombarded moderators with complaints." .... Oh my

@Skillmon What's the point of hiding anything now? And what is it exactly that you'd like to hide? Just to take an example, I might have some very strong opinion about it, but it may be different from that of my prior opponent.

@user0 it's not about hiding it now, it's about future handling of comments which might be offensive to some (so are close to the border between deletion and keeping).

@user0 When I check and possibly remove a flagged comment (off-topic or offensive or abusive) I try to clean related comments since they may be obsolete after removal as well. But, please don't flag a single message and expect us to find out what to delete and what not (in a thread of 30+ comments). Simply flag each single message if you like. (generally meant, not at you)

So no, I don't delete a thread that I have to investigate just because of a single flag. users can flag every single comment if they would like to push the work

@Skillmon Do you mean future hiding of future comments or of any comments?

@StefanKottwitz Oh, I don't have any to flag anything now. But thanks for letting me know.

@user0 yes, just in general

@user0 as sometimes there are flags, lets say, fictional, "I got this and that unfair all the time" - how on earth can we check everything... in a bulk thing, a user could flag each single thing to lead the way

@user0 I mean that if there were other flagged messages in chat in the future and they were not full fledged bad, then they'd get hidden.

Btw., I guess, there is no point in asking whether tex.stackexchange.com/users/36296/samcarter got too many flags... Some time ago, I observed that the user was argumentative once in a while, but not tremendously argumentative that would warrant a ban.

@user0 True, as it's not productive to talk about users. Issues can be raised on TeX - LaTeX Meta.

@user0 Asking is not forbidden, but don't expect answers; and if there are answers, they would contain rumours, most likely, not profound knowledge. And it is most times better to talk to persons, not about them (if possible)

@ChristianHupfer I'd love to ping her and get a response from her :(

@ChristianHupfer Thank you, I'm glad that you said it. (Though it's a bit late; that would have been exactly my point in an earlier, now-closed discussion.)

going to bed now. Good night fellow users!

@Skillmon I know her real name, but I have no way to contact her ... some others do have know her mail address, however...

@Skillmon Thx and good night!

@Skillmon Good night

By the way, when the design of tex.se changed, some folks protested, saying they'd move to another site (I recall them mentioning something supported by TUG). So, has any viable alternative arised since then?

@user0 I have not seen percusse since them, and consider this a dramatic loss. (However, I cannot say for sure that there is a relation between these events, but the timing may suggest such a relation.)

there are German and French alternative sites (even a br Portuguese start) but for native language user support

@user0 There is a German speaking alternative to TeX.SX, but that existed before the design change: texwelt.de/wissen

@HenriMenke I know texwelt.de. I took a look at it right now and saw more folks from here than it was the case before the design change. So, apparently, some movement has arisen :-).

@user0 not much sense to create a competing site from zero as all depends on users asking and users go to the site google shows first, so the pro's site, and most users are happy here anyway

@user0 It's hard to tell. I contribute over there pretty much since the beginning (registered user number 7), but the frequency of questions hasn't changed that much.

@StefanKottwitz will you pop into TL.

@ChristianHupfer Concerning the particular banned user, taking a look at her latest (still available) activity led me nowhere.

@user0 Would you please refrain from going after other users here, in particular those who cannot react? This is not the purpose of the chat.

@user0 nothing to see, leading to nowhere is ok for today, let's leave it today

@marmot @StefanKottwitz Nothing personal. I just wished to see the kind of behavior that finally triggered the ban which could be a little bit more specific than that of "Your Common Sense" - if that was supposed to be a hint at all. I'm not interested in "going after other users" or after her specifically or judging on how harmful her action were in reality. So, I apologize and wish everyone good morning/afternoon/evening/night.

@user0 Sorry, I must have misread your statement. I'd like to apologize for that. (Led me to nowhere can mean different things.)

@Skillmon I understand your point of view, many software versions are numbered in this way. This is not what has historically been done by the designers of TikZ. This change in numbering needs to be clarified.

@AndréC actually this scheme is bound to change once I consider the bundle stable enough (which would be after I wrote my next thesis sized report with it). Then the version numbers will all jump to 1.0.0 and from there on bugfixes will be increment of last number, features increment the second number and major changes the first number (unlikely to happen). Repackaging will not occur since its not released package wise but will be pulled directly from its git, so no alphabetic appends.

morning

@JosephWright just trying again the specialformats fields. How (and where) I'm supposed to add an entry? Currently I created a config with specialformats.latex.luahbtex={binary="luahbtex",format="luahblatex"} checkengines={"luahbtex"} , but it starts the normal luatex.

@UlrikeFischer Strange: you shouldn't even need an entry for that, as it matches the <thing>tex model

@UlrikeFischer I had it working with luatex53: I'll double check in a second

I have a problem with a tabular table inside a multicolumn inside a longtable. The columns in the longtable are all defined as p (vertically top bound) but when the tabular gets more rows inside a single longtable row, the other columns move to being vertically centered

for some reason

@JosephWright I think I doing something dump. I'm a bit confused regarding engine/binary/format names.

@UlrikeFischer Do you have a .fmt file called luahblatex.fmt? If you do luahbtex --ini latex.ltx it will start off being called latex.fmt

@JosephWright I can run luahbtex --fmt=luahblatex.fmt on the command line without problems. Also I can see that l3build is calling luatex and not luahbtex. I got it working by using specialformats.latex.blubblub = ... and checkengines={"blubblub"}, so I think somewhere in the code l3build is matching luahbtex with luatex.

@UlrikeFischer There is some code to match the .tlg files (comes from luajittex), but it should be running the right engine ... I'll have a look

@JosephWright arab1.blubblub.fc looks actually quite nice - perhaps I keep this ;-). I get quite a lot failures due to different line breaks - the log lines are much shorter in luahbtex.

@UlrikeFischer Not just 'out by one'?

@UlrikeFischer That's weird: is it not respecting the normal settings

@JosephWright no, much more. e.g. with luahbtex "with language dflt" in on a new line. In one of the arab tests, the tlg has line length of 136, but luahbtex breaks at 88. But don't forget that the "engine" is called blubblub. Not quite sure how l3build should know that this is a luatex-like engine ...

@UlrikeFischer Working on it: it's tricky ...

@UlrikeFischer Right, perhaps I need to rethink on the engine 'co-mangling'. It's there as luatex and luajittex should give the same results, so was meant to avoid needing two sets of .tlg files. However, that doesn't hold for luahbtex, and I'm not sure it really does for luatex53 (rounding changes to glue setting). So perhaps I should just strip out the entire idea, and treat these all strictly separately?

Good morning everybody and good work.

@UlrikeFischer If I drop the idea of a separate engine and binary setting, a lot of the code gets less complicatd. Things should then 'just work'.

@JosephWright I'm not sure if the tlg are the problem (sometimes pdftex and luatex can share a tlg, sometimes not ...), but I assume that l3build contains engine specific stuff to set line length, suppress fields in the pdf, set the compresslevel etc. So it will have to know which commands to use for the "blubblub" engine.

@Raaja Thank you very much for reject the edit and thank you to Stefan Pinnow.

@UlrikeFischer There is that: could go into the specialformats table. The Lua code only needs to know when to use the extra LuaTeX normalisation: everything else is macro layer

Hi, dumb question: I was just looking at an old document, and thought there must be a better way to smash a superscript than C^{T\!\!\!}, as in $\begin{bmatrix} A & C \\ C^{T\!\!\!} & B \end{bmatrix}$?

@UlrikeFischer Hi and good chat

@UlrikeFischer My point was that with LuaJITTeX, sharing always works with LuaTeX, so I special-cased it. But that breaks down with multiple LuaTeX-like engines.

(I usually left-aligned matrix* environments for that, but I was curious)

@JosephWright yes, I think breaking this up is a better idea. You still could write a "luatex-defaults" table for the various settings and reuse it for other luatex-like binaries.

@UlrikeFischer Perhaps a luatex_like table?

@JosephWright sounds like a good plan something like specialformats.latex.luatex={settings=luatex_like, ...}.

@UlrikeFischer I was thinking just luatex_like = {"luatex", "luajittex", "luahbtex", "luatex53"}. First, let me see if I can get the basics working without any odd names!

@JosephWright what is odd in "blubblub" compared to "zzzzzzzz"? ;-) But if you do such a luatex-like table, won't you need a pdftex_like, xetex_like table too?

@UlrikeFischer No, because only LuaTeX has the extra normalisation. The only other odd setting is the --no-pdf one for XeTeX, and that's only a performance thing so it's not vital

@JosephWright Is there a possibility that luatex needs somewhere a different normalisation setting as luatex53 or luahbtex?

@UlrikeFischer I don't think so: all of that code is to do with LuaTeX's deliberate changes

@UlrikeFischer Try very latest check-in

@JosephWright with which specialformats name? specialformats.latex.luahbtex?

@UlrikeFischer Doesn't need to be in specialformats, assuming you call the format file luahblatex.fmt: should just work (does for me)

@JosephWright I get again an error about a "boolean enginename".

@UlrikeFischer Really? It's working away happily here ...

@UlrikeFischer I've taken out enginename entirely ...

@JosephWright grep shows various enginenames so I must have messed up something.

@Magisch no the other columns will be top aligned but the tabular is centred by default use \begin{tabular}[t]

@UlrikeFischer Hmm, I'll look again

@Sebastiano ;) welcome

@Skillmon This is the classic way of numbering software. If your version is 1.9.9 how will you number the following 2.0.0 or 1.10.0?

@DavidCarlisle The tabular is defined as \begin{tabular}{t{0.03\textwidth}t{0.035\textwidth}t{0.222\textwidth}@{}}%

where t is \newcolumntype{t}[1]{>{\raggedright\arraybackslash}p{#1}}

Also, for some reason the new tabular inserts an extraneous whitespace into the row before beginning, so it's like it's a whole row of the table offset. fudging it with \vspace{-\baselineskip} works but looks like a massive antipattern

Also, remembering the different conventions between longtable, tabular and tabularx is a little daunting

@AndréC the latter. Features remain second number. Only bigger changes change the primary versions.

especially since I have a longtable containing a tabularx containing 2 tabular containing 2 tabularx

@Raaja Again welcome :-) also to you. My English language is very bad: you must remember always :-)

@UlrikeFischer Hopefully fixed now

@Magisch yes as I guessed, so it is vertically centred, use ` \begin{tabular}[t]`

@Magisch there really are no differences at that level, both longtable and tabularx use a normal tabular to do all the actual typesetting,

@Magisch I would say 99% of the times people use tabularx they would be better not using it.

@JosephWright 17/17 tests passed with only checkengines={"luahbtex"} in the config ;-). Now I only need to think about how to setup things so that travis doesn't complain ...

@UlrikeFischer Good

@UlrikeFischer That's for luaotfload?

@UlrikeFischer Won't that have to wait until there's a TL version of LuaHBTeX? Or you have to add installation manually to the script, e.g. by pulling from the GitHub repo and building/installing up-front

@DavidCarlisle tabularx protects me from my own failure to properly calculate column lenghts

more then anything

@JosephWright I can use the binaries from w32tex.org, there are easy to install locally. I naturally can't run tests on travis, so I need to setup the build.lua so that the tests with luahbtex are not triggered there - should not be difficult - I can simply check if my mail address is known (I pull it it here with a local file).

@Sebastiano dont worry, my English is not so good either ;)

@CarLaTeX Yes. Sadly I don't understand a word of the four Chinese words :)

@UlrikeFischer more portably you can test if os.getenv('TRAVIS') is 'true' in the build.lua

I just noticed btw

You wrote both the tabularx and the longtable package?

@Magisch someone much younger and more naive, who had the same name...

Is that why there's a "Died" symbol next to your name

@Magisch have ctan killed me again ??? that happened before I had better check....

@Magisch I see no symbol at ctan.org/pkg/tabularx

Unless I'm reading that wrong

@Magisch oh no that's in the document, that is just the standard latex footnote marker to a note at the bottom of the page

ctan.org/pkg/longtable?lang=de links to -> mirrors.ctan.org/macros/latex/required/tools/longtable.pdf

@Magisch (and yes it is me:-)

Oooh so I am reading that wrong

@Magisch yes I was worried because ctan do mark authors who have died see for example Sebastian here ctan.org/pkg/hyperref and a while ago the data got corrupted and they were marking me as died on all my packages for a few days, I got some concerned emails, people never know how to email someone and ask if they are dead....

oh

It's not latex that is making me tear my hair out, it's the fact that I have to use it as an arbitrary template engine

to basically capture all use cases & iterate upon the MS access reporting system

@Magisch That symbol for 'deceased' is usually not typeset as a superscript, but rather like 'John R. Tolkien (†)' (or using similar symbols)

@ChristianHupfer I know he's deceased as he's in the field next door:-)

@DavidCarlisle You told me once, already ;-)

@ChristianHupfer what the same conversation comes up more than once in this chat? Shocking.

@Skillmon This is not the most common choice in software development. Firefox and WordPress have chosen short development cycles. There is a new version every 4 months. Thus, there are no longer any major changes in the software, but a succession of small changes. WordPress has been upgraded from version 3.9 to 4.0. make.wordpress.org/core/handbook/about/release-cycle/…

@UlrikeFischer Are we good to update l3build on CTAN?

@DavidCarlisle No, be we talked about Tolkien (and his grave) once some years ago and I know since then that his buried in Oxford near your working place

@AndréC semver.org/spec/v2.0.0.html

@AndréC define "common" a few systems are following some kind of rolling update system but I would guess the vast majority of software is versioned in more discrete versions

@AndréC that's a nice dream :)

@AndréC I don't think you can say software must always be backward-compatible. It's very hard never to break stuff, simply because the result is often a fork which does the same thing. Certainly the team try to keep stuff compatible, as I do for my own code.

@AndréC I'd like that. On the other hand I'm a software developer by trade and that never happens or works outside of very few circumstances

@AndréC The team have brought in a 'rollback' mechanism exactly to cover this: the best one can do sometimes is to have a 'use the frozen older code' approach

@Magisch Since the software in the TeX universe is built from the TeX kernel, which is never modified, LaTeX must be the example for all other software and always be backward compatible.

@yo' It was Donald Knuth's dream. It's mine too. :-)

@AndréC we all do often dream :) Note that so far I keep a working TeXLive for each year since I started my job in publishing industry :)

@yo' How do you get multiple functional versions on the same computer?

@AndréC luatex changes with incompatible changes (adding/removing/renaming primitives each release, you can't use the current latex sources with a tex system from 1982 as they use etex extensions, tex itself is completely stable but essentially no one uses it they use pftex etc

@AndréC in linux this is an easy trick with PATH and stuff (I'm not able to show it now as I'm not on my computer)

@AndréC that is trivial each texlive installs into a new directory and you just need to set the path to pick up the right version.

@DavidCarlisle I never liked LuaLateX and you give me a good reason to never use it. :-)

@AndréC actually you don't need to set the path most of the time, you can simply go /usr/local/texlive/2016/bin/x86_64-cygwin/pdflatex small2e

@AndréC same is true to a lesser extent with xetex, and at some point luatex might be the only practical choice.

@AndréC no that is false, you can not build latex on Knuth's tex.

@DavidCarlisle And why is that?

@AndréC because etex added useful functionality in 1993 and by 2011 it was just getting embarrassing thinking of reasons why we were not using it. Making it work in classic tex is hard (not feasible in some cases) and strains the resources for testing and has no practical advantages as no one uses classic tex with latex: all distributed latex formats have been built with etex since last century.

@JosephWright Thank you for writing texdev.net/2011/07/05/expansion-using-romannumeral

@DavidCarlisle I thought etex was just giving TeX more registry. If I understand correctly, it does more than that. What does etex really do, what does it bring that is so vital?

@JosephWright ...but one question on it. Why does \def\testa{\testb}\def\testb{\testc\testc}\def\testc{abc}\demo{\testa} expand to abc\testc rather than abcabc? Given that there are no spaces in the argument.

@AndréC w3ll 32 thousand more registers is useful, also \numexpr 1+1, \dimexpr, \scantokens, \marks, ....

@AndréC The full story can be found here: texdoc.net/pkg/etex

@StevenB.Segletes it stops expanding at the first non-space

@DavidCarlisle I thought Joseph said it expands until reaching the first space. ("TeX always looks for an optional space to gobble after an integer"). So, if it stops expanding at the first non-space, that means that once the first \testc gets expanded to abc, the finding of the a terminates the \romannumeral expansion??

@DavidCarlisle I see now. An "unexpandable token" is the a found in the first instance of \testc.

@AndréC try for example your answer here which would not work in classic tex

For some reason \begin{tabular}[t]{t{0.03\textwidth}t{0.035\textwidth}t{0.222\textwidth}@{}}% continues to generate a full line of whitespace before it begins

I have no clue how to make it not do that

@DavidCarlisle A very good example -- with nice answers by some blokes currently present in chat ;-)

@DavidCarlisle TeX and LaTeX are so complicated that I ask myself the following question: who will replace the pioneers of LaTeX and TeX? Won't the complexity of the system be too great for newcomers?

@AndréC is latex any more complicated than the source of chrome, or the linux kernel, or farming simulator 2019 o r....

@Magisch no, it doesn't, you may get a line of white space, but make an example and post a question on site, tabular itself makes no space outside its box.

@StevenB.Segletes Yes, that's the case: first non-expandable token, if it's a space it's consumed

@DavidCarlisle These softwares are maintained by professional developers who spend their days working on them.

@JosephWright FYI @DavidCarlisle jumped in and answered my questions. I liked his description: "first non-expandable token, if it's a space it's consumed"

@DavidCarlisle @JosephWright @Magisch @AlanMunn @yo' @Skillmon I have to leave, have a good day and see you soon. Nice to have talked to you all.

@DavidCarlisle that's an idea too thanks.

@JosephWright I could test this evening @DavidCarlisle's code for upload target.

@JouleV Imagine what I can understand, lol!

@AlanMunn I didn't know that Peter Breitenlohner did such important things for LaTeX. (He did really important stuff such as the Breitenlohner-Freedman bound, and in his spare time he volunteered to give lectures on supersymmetry and group theory... ;-)

@marmot he did almost all the main development in etex and for years much of the texlive maintenance as well

@DavidCarlisle I see. That's why he was always that busy.

@PhelypeOleinik ^^^^^

@marmot Ooh, clever marmots :)

@marmot I'm waiting for my "The Star Diaries" to arrive :D

Quack!

@DavidCarlisle oh no

@DavidCarlisle you are mean

@DavidCarlisle oh

@DavidCarlisle ooh

@PauloCereda 5776

@DavidCarlisle ooh a counter

@marmot I had no idea he was a physicist. This is kind of like the people who only know Chomsky from his political writings and are surprised when they find out the same guy created modern linguistics. Vice versa not so much, since I think most linguists know about both.

@AlanMunn Before he started smoking, he could climb 1000 m in below an hour on hikes, rumors say.

@marmot One wonders why he started smoking then.

@DavidCarlisle is there a sensible way to I check in the build.lua if luahbtex is in the path? I need to exclude its tests when running on tl18 or miktex.

@PhelypeOleinik, i just tested your fix (from yesterday) for row sep. it solve my problem asked in my question tex.stackexchange.com/questions/469954/…? please, can be so kind and provide an answer on this question. I thing that with your answer more people will be informed about your fix. i hope that it will be in near features part of the next TikZ version.

@UlrikeFischer errorlevel = os.execute("luahbtex --version")?

@Zarko The fix was actually made by Henri Menke, in the official PGF repository. I just copied it here. But yes, it will porbably be in a future TikZ release.

@Zarko But if you want I can post an answer :)

@UlrikeFischer just try to os.execute luahbtex --version and pick up the error or not?

@JosephWright ^^ oh you said that already

@JosephWright would be an idea, but probably doing kpsewhich whatever-option-it-was luahblatex.fmt is easier. (I found a bug in the current luahbtex-harf or luahbtex on my first test ;-().

@PhelypeOleinik, i didn't know that @HenriMenke is the author of this fix. Many thanks to him for this valuable work! Anyway, your answer is welcome!

@JosephWright Hm. Using kpse doesn't work, at least not with the kpse library, it doesn't find the format, there is no equivalent for the --engine key.

@Zarko Done :)

@UlrikeFischer Yay!

@UlrikeFischer So can I update CTAN?

@PhelypeOleinik, thank you! i accept it :-)

@AlanMunn To spend less time in the office? (If you do day hikes in 2 hours, you will end up staying too much in your office.)

@JosephWright sorry got a bit distracted with this tests. I'm just building the zips and will then try to upload.

I may be late to the party, but I've recently picked up on the change in maintainership of TikZ/PGF. Is there a summary of what's happened and what the future holds available somewhere?

@LoopSpace don't worry, if it all goes wrong, we've still got picture mode.

@DavidCarlisle I knew you'd be kept in the picture.

@LoopSpace did you see my latest high scoring tikz-pgf answer?

@LoopSpace I guess @HenriMenke is best placed to answer your actual question

@LoopSpace l3draw ;)

@JosephWright Yes, I've been out of the loop on that one as well. How're things going with it?

@DavidCarlisle Is that the same as your first high scoring [tikz-pgf] answer?

@LoopSpace Not much recently, I want to check some performance issues (on the back of my FPU work)

(How do you get fancy tags in chat? I've been away so long they've upgraded the markdown without me noticing.)

@LoopSpace judge for yourself

@LoopSpace [tag:tikz-pgf] (same as always:-)

@LoopSpace [tag:foo]

@DavidCarlisle I'm too lazy to search - can you link to it?

@LoopSpace [like this](https://tex.stackexchange.com/a/473314/1090)

I'm also too lazy to spot that you've already linked to it.

@DavidCarlisle I'm disappointed - I appear to be able to understand the commands. I thought it was axiomatic that all your answers had to be obfuscated.

@JosephWright Thanks. I really didn't know that that was possible

@LoopSpace no you just though that as tikz code naturally looks obfuscated, but that's classic and clear picture mode.

@JosephWright Well, it looked promising. But I know full well that other_things get in the way of TeX.

@LoopSpace I recently joined the development team and am trying to fix all the open bugs. As to what the future holds I have no idea. The PGF/TikZ project is pretty much abandoned because Till Tantau has not responded to any communication for about 3 years. The other project owner Christian Feuersänger is quite busy in real life, so he only ever jumps in if there is a severe bug.

@LoopSpace I have offered to revive PGF/TikZ by moving the project from SourceForge to GitHub, establish a good automated test coverage and modernize the macros to make more use of e-TeX features. However, this offer was declined with the rationale that PGF/TikZ is “Till's baby” and we must not take it away from him. So far I can accept that but sooner or later it will become unusable and has to be forked or “taken away” from him.

@LoopSpace Because I don't have any authority over the project I also don't feel too inclined to contribute any new features. So far my contributions were focused on responding to new bugs and trying to fix old ones.

@DavidCarlisle if l3build upload -debug gives me back a description field with ogramming.\\n The bundle is based in it I should probably be worried about the newlines?

@HenriMenke I see. That sounds like you're doing a sterling job holding the fort. I hope it gets resolved before it truly stagnates. If you ever do fork it, or are able to take over properly, do let me know.

@UlrikeFischer really:( with two \\n not \n what does the .curlopt file look like?

@LoopSpace There is also the plan to add the information on what libraries are to be used in the examples in the pgfmanual....

@DavidCarlisle where do I find it?

@UlrikeFischer next to the zip file hopefully

@UlrikeFischer it should look something like

$ more test-ctan.curlopt

form="announcement=test\nannouncement\nline 3"
form="author=me"
form="ctanPath=/a/b/c"
form="description=An interesting package\ndescribed on several lines\nwith \"dou
ble\" and 'single' quotes."
form="[email protected]"
form="license=lppl"
form="note=a note\nto ctan"
form="pkg=test"
form="summary=ZZzzzz"
form="update=true"
form="uploader=me"
form="version=0"
form="[email protected];filename=test-ctan.zip"
url=httpbin.org/post

@DavidCarlisle there is no curlopt file. Does it get created if one use the debug option?

@UlrikeFischer are you using the version from the curl-file branch?

@DavidCarlisle no. Sounds as if I need to merge with @JosephWright changes.

@DavidCarlisle ah better: mming. \n The bundle i ;-)

@UlrikeFischer phew:-) merging should be easy as the only changes on my branch are to l3build-upload.lua (or simply use the master branch to build, and the ctan-file branch to upload)

@DavidCarlisle yes, was easy. I copied everything in the main texlive branch - as l3build changes the texmfcnf variable, using some local tree doesn't really work. Btw: I forgot the announcement file and got a message that I could stop with "ctrl D" which didn't work.

@zylitoL Welcome to the chat!

@UlrikeFischer hmm three returns also works, I should modify the ^D suggestion if it doesn't work in windows (does ^Z work there)

@DavidCarlisle and now for the real stuff. What does that mean:

Warning: error trying read config from the 'lualibs-ctan.curloptvalidate' file
curl: no URL specified!

@UlrikeFischer sigh it means I only tested with the http test site, sorry, actually that's a bit of a pain

ctan uses two url .../submit and .../submit/validate so the validation step on the commandline can simply tack validate on to the end of the command and it works, but here the submit url is embeded in the url line of the .curlopt file so the validate step needs a different file. (or need to edit the file in between) give me a minute...

@UlrikeFischer @JosephWright curl has a manual.... OK it seems I don't need to specify all the options in the config file I can keep the url out of the config file then supply the (validate or submit) url on the commandline,

$ git commit -m "keep URL on commandline so can use validate or submit"
[curl-file 33168c9] keep URL on commandline so can use validate or submit
 1 file changed, 7 insertions(+), 7 deletions(-)

@UlrikeFischer ^^^ sorry about that:-)

@DavidCarlisle I still get an error about the curloptvalidate-file. And in the curlopt file I now have a stray "https://httpbin.org/post" at the end. (And I hope I copied the right file ;-).

@UlrikeFischer I may have messed up, let me try again to test on windows and with and without --debug

@UlrikeFischer update coming up ...

@UlrikeFischer git commit -m "add url at the right place"

@DavidCarlisle after a small hickup as I had one license entry still wrong, two uploads were successfull. ;-)

@UlrikeFischer If Petra complains that all the newlines appear as \n I'll cry....

@UlrikeFischer ^^^ This seems to be a bug in pgfplots which was uncovered by the PGF update.

@UlrikeFischer See the answer I posted.

@HenriMenke quite possible, I rather carefully didn't say where the bug is ;-) (the command name pointed to pgfplots, but on the other side pgf has changed).

@Sebastiano how do you do?