@PauloCereda -- Here's an island and a little bird for you to learn about: thehill.com/changing-america/sustainability/environment/…

@samcarter_looks_forward_TUG'21 Yes, just munch together the slides and go for PowerPoint-karaoke (powered by beamer)!

@Skillmon Speaking of beamer ... lots to do there :)

@barbarabeeton ooh cool, thanks!

@PauloCereda could you test something for me?

@JosephWright I see you were busy - just got all the github mails :)

@samcarter_looks_forward_TUG'21 Yes: looked at an issue that came up from team discussions, though I'd look over a few other issues. That \emph one is ... tricky

@Skillmon I bet that would be fun

@JosephWright It is just such a commonly used macro that it would probably impact many users if changed now. Maybe we can add a line to the doc stating that it works differently in beamer?

@JosephWright what is tricky with emph?

@UlrikeFischer github.com/josephwright/beamer/issues/623

@UlrikeFischer AS @DavidCarlisle says, probably the beamer behaviour is wrong, but we can't really change it (There are various things that beamer, er, does differently)

@JosephWright yes looks so.

@samcarter_looks_forward_TUG'21 can you test something on linux for me?

@UlrikeFischer At least, not unless I could find the time to do a real overhaul for v4.0, with all the rollback required: I'm not sure that works as I'd want to alter some of the basics, and as I said, reallly it's only worth it if we ever sort styling properly

@UlrikeFischer I can try

Is it only me, or did they change the way moderators are shown on meta? Now they get displayed with an ugly blue rectangle around the diamond and the abbreviation Mod.

@Skillmon yes, that was a feature request because staff members where not recognisable (many of which have 1 rep, so it was hard to know if a troll or staff)

@samcarter_looks_forward_TUG'21 so staff now gets displayed with an ugly blue rectangle with "Staff" inside?

@Skillmon orange, but yes

@samcarter_looks_forward_TUG'21 why do they get orange and the poor mods have to live with cold blue?

@Skillmon because blue is the most beautiful colour there can be :)

@samcarter_looks_forward_TUG'21 especially on a site which uses a bright warm orange for everything else... :)

(and this comes from someone who considers blue his favourite colour)

@Skillmon a tiny problem solvable by a user script :)

@Skillmon good choice!

@samcarter_looks_forward_TUG'21 I simply want to know if latex2nemeth works on linux. You only need a small helloworld.tex, and then should run first pdflatex helloworld followed by latex2nemeth helloworld.tex helloworld.aux

@UlrikeFischer I get

latex2nemeth test.tex test.aux
Parsing file: test.tex
[main] INFO com.latex2nemeth.io.DataDumper - Creating test0.nemeth

@UlrikeFischer seems to work. This is the output for "hello world!": ⠓⠑⠇⠇⠕ ⠺⠕⠗⠇⠙⠖

@UlrikeFischer terminal output is the same for me as @samcarter_looks_forward_TUG'21's

@Skillmon file output is the same for me as @Skillmon :)

@Skillmon thanks. So only windows is wrong ;-(. @samcarter_looks_forward_TUG'21

@UlrikeFischer windows is always wrong :)

@samcarter_looks_forward_TUG'21 what she says!

@samcarter_looks_forward_TUG'21 well actually it is not windows, but the package. It gave the jar the wrong name, relying on the sh-file to start java, but it isn't there on windows (arara does it correctly and that works).

@UlrikeFischer That does not conflict with windows being wrong. Both can be wrong :)

@samcarter_looks_forward_TUG'21 I'd say that Windows is wrong here for not expecting sensible names. :)

@barbarabeeton Maybe the meta question linked above would be a good place to raise the problem with community ads for conferences?

@MarcelKrüger I've started on a GitHub Actions setup for siunitx (github.com/josephwright/siunitx/tree/actions), but the caching doesn't seem to work: could you take a look when you have a chance

@MarcelKrüger Also, where should I look to get deployment set up for tags? I've not got it on Travis-CI anyway, but it would be nice to have it working :)

@MarcelKrüger Scratch that: I see caching is working

@MarcelKrüger So just need to do deployment :)

@JosephWright Depends on how urgent it is. At the end of the week I'm planning to setup GH Action with deployments for luaotfload so then you can just copy it from there, otherwise you can just change the l3build command to l3build ctan and add github.com/marketplace/actions/github-releases as additional step.

@UlrikeFischer sure!

@UlrikeFischer oh sorry, I was late to the party... :(

@UlrikeFischer oh it's that Java program... want me to take a closer look, programming-wise? Maybe I could find what's going on...

@JosephWright follow longtable's philosophy: everything is user error. :)

@MarcelKrüger No urgency - I want to adjust how things run a bit first (I get why you've used a shared approach to teh install.sh script, but that means I can't jsut checkout and run on a local VM: I want to be able to do that)

@PauloCereda I have to go in a minute, I managed to get latex2nemeth working by renaming the .jar to latex2nemeth.jar. But now it fails because it can't find "\encoding\nemeth.json" (which should be included in the jar). On unix it works, I wonder if you have an idea what is wrong with the java code here).

@UlrikeFischer I will investigate

@MarcelKrüger So far it's all worked smoothly, so I'm quite happy: like I said, Travis-CI for some reason not doing deployment anyway, so there's no major issue in waiting. (As siunitx is small, I already use l3build ctan -q -H rather than doing separate check and doc tests)

@UlrikeFischer I have something in mind, will write to you later.

@MarcelKrüger I think we will need that for the LaTeX2e/3 repos - they take ages to build locally, so it would be good to stick to a deployment approach

I'm trying to figure out why putting a label in longtable produces a multiply defined label warning. I've read the following in the longtable manual:

My example code is in gist.github.com/a84198ecedc6f39c3ea59064e4a647f5

But I've tried putting the \labelmacro all over the place, but it still gives that warning.

If this is suitable for the site, I can post it there. But I think it might be something obvious. And there are similar questions there already.

From the documentation, it looks like anywhere after \endhead should work. Because that's the bit that gets repeated.

@FaheemMitha No! You put it in the \endfoot part which is also repeated. That's why the docs say “should not be used in a heading that may appear more than once”

@TeXnician Oh. I guess I don't understand the syntax properly.

@FaheemMitha Well, obviously the endfoot part is repeated more than once otherwise you would not get your “continued” message more than once…

@TeXnician Oh. Right. Though I'm unclear where the startfoot is.

@FaheemMitha startfoot?

@TeXnician Where does the header end, and the footer start? And isn't there supposed to be something in between?

@FaheemMitha \endhead ends the header, then something else starts and \endfoot marks that this “something else” was actually the footer, so from endhead to endfoot is the footer. And I don't know why there should be something in between…

@TeXnician Oh, from endhead to endfoot is the footer? That wasn't clear to me. Something in between would be the body of the table?

@FaheemMitha In this example that is the footer. You should read up on the commands to actually get the rules. I'm only talking about your example here. And it wouldn't be useful in a longtable to put the body before the repeating parts because that's supposed to be long. So you specify the header and footer upfront.

@TeXnician OK. This wasn't clear to me. And I did look at the manual.

"Section 3: Captions and Headings" appears to have all the documentation available on the subject. And most of the discussion of headers and footers is in the first para.

And from that it isn't clear that footer immediately follows the header. I think it would be clearer if there was a start command, or macro.

@FaheemMitha Yes, and section 3 starts with “At the start of the table” which answers one of your questions right at the beginning. Admittedly, it's very comprehensive, but there are tons of longtable examples on this site.

@FaheemMitha Well, actually it does not have to be clear. It says at the start of the table. So when you have specified the header you may specify anything else as long as it's still before the body. And there is no ordering, you can swap header and footer specification so that the code for the header follows footer. Hence, I said I only explained your example, nothing more.

@TeXnician Did you mean to write "it's very comprehensive"?

@TeXnician There's no ordering? I thought the header came before the footer.

You mean you could end the footer and then end the header?

@FaheemMitha Ah, not really. Wrong translation in my head. Let's say “it covers everything (but does so on a more abstract level)”.

@TeXnician I think it's quite unclear. But I've never managed to convince anyone here of any deficiencies, ever. So I'm not sure why I continue to try.

I guess I could open an issue, but I'm sure it will be ignored.

@FaheemMitha Yes, of course. You can try in your example.

@FaheemMitha That's why I explicitly told you In this example and to read up for the more general stuff.

Well, that's unexpected.

@TeXnician Read what? I already read that section. Is there some other documentation I'm missing?

@FaheemMitha You could ask @DavidCarlisle to clarify stuff but as I said, most people will look at the tons of exmples on this site rather than reading the docs.

@TeXnician I thought the documentation was there to be read. Isn't that why it exists?

I suppose the experts would simply look at the code to understand what's going on, but that's not a good option for everyone.

@FaheemMitha The documentation is quite clear about the termination part. Of course, it's technical but you already have some TeX background. If you don't want it to be too technical you have very good chance to find proper examples on TeX.SX that should make things clear.

I confirmed that flipping those gives the same result.

@TeXnician The termination part?

@FaheemMitha Right, most people do “5 secs of google”, find something from TeX.SX, copy & paste it into their editor and use it without understanding anything. But this surely isn't about “most people” because that would exclude reading docs ;)

@FaheemMitha The docs mention that header/footer etc. are code section terminated by commands like \endfooter, so yes, termination.

The most non-obvious thing here, is that the footer section immediately follows the header section. And that neither of them actually have a start marked. It might also be worth observing that they can be flipped, which is certainly not something that I would have expected.

@TeXnician That part, yes. But that's only a part of the story.

@FaheemMitha Why?

@TeXnician See above, starting with "The most non-obvious thing".

@TeXnician Sometimes I do the "5 secs of google" too. But then you don't learn anything. And often might have lurking bugs in whatever you've blindly copied and pasted. It's happened to me before.

@FaheemMitha Well, I think that it's obvious, you don't. If in doubt, ask the author of the package. In my view, it's a matter of reading the docs carefully (everything I have told you is from the docs and some times of me using the package), you don't agree. After all, these are two views on the matter.

The first paragraph of section 3 is somewhat the algorithmic description of how to obtain a header, a footer, a different first head and a different last foot. And it also says that the footer is specified at the start of the table.

@FaheemMitha Of course there's always a tradeoff. But most LaTeX users I know are doing exactly that so they would not even wonder how the documentation is written ;)

@TeXnician Yes, I know. I read it too. But it doesn't tell me everything I need to know.

I'd start by saying that the header and footer for the table are specified at the start of the table. But then I wouldn't quite know what to say, because I don't understand how \endhead and \endfoot are implemented, or how they work.

It would seem reasonable to mention that the sections for the header and the footer follow each other immediately, so that is the entirety of the table before the body is written. (Something like that.)

And the manual says:

From that one would not guess that one could swap the matter for the header and the footer. At least, I wouldn't.

@FaheemMitha “At the start of the table one may specify lines which are to appear at the top of every page … [explanations on head] … There are also \endfoot and \endlastfoot commands which are used in the same way (at the start of the table) … [explanations on foot]”. Well, it starts with these pieces. And the at the start of the table part is obviously what says it comes before the body.

As I said, I am able to read everything from this, but you could of course make a pull request with better docs at github.com/davidcarlisle/dpctex ;)

@TeXnician Yes, that's all well and good. But it leaves a fair amount of ambiguity, which gives scope to confusion. I'm now wondering how \endhead and \endfoot are implemented. Specifically, how do they know how not to include all the stuff from the start of the table, since there is no corresponding start command?

@FaheemMitha I suppose that the \endfoot reads either from the start of the table or from the last \end... something found, like \endhead.

@FaheemMitha Then you will simply have to open the sources ;)

@Rmano Hmm.

Looks like the definitions are not so complicated. They are:

\def\endfirsthead{\LT@end@hd@ft\LT@firsthead}
\def\endhead{\LT@end@hd@ft\LT@head}
\def\endfoot{\LT@end@hd@ft\LT@foot}
\def\endlastfoot{\LT@end@hd@ft\LT@lastfoot}

Looks like there's an explanation of sorts in "9.12 Commands for the table head and foot".

@FaheemMitha longtable starts collecting its contents inside chunks, and only after a chunk is completely read it decides what to do with it. So at \begin{longtable} it immediately starts reading the first chunk of rows. If during that process it finds an \end(head|firsthead|foot|lastfoot) it finishes that chunk and stores the contents inside a special register (\LT@(head|firsthead|foot|lastfoot)). Those registers are then used during the output routine to place the respective head/foot.

@Skillmon I see. Thank you. That's informative.

@Skillmon Oh, that's what those macros mean.

Is there some reason it can't work with begin as well as end markers?

@FaheemMitha simplicity, and the headers and footers have to be read before the first output is triggered. So this method is the shortest syntax which allows for all the possible errors to be detected.

@FaheemMitha the begin markers are implicitly given. The first one is \begin{longtable}, and all other begin markers are given implicitly by the other end markers.

@Skillmon So any time it sees a \endsomething it finishes up that chunk and stores it away? And then proceeds with the next chunk?

@FaheemMitha correct.

@Skillmon Explicit is better than implicit. That's in the "Tao of Python".

@Skillmon Are you saying that those header and footer rows could appear in the body of the table too?

@FaheemMitha no, they have to be the first things inside longtable.

@Skillmon OK.

@FaheemMitha longtable will throw an error if it detects that you messed this up. The error will read "Longtable head or foot not at start of table" and the additional help will be "Increase LTchunksize".

@Skillmon Yes, I saw that bit of code.

Thank you for the explanations.

@Skillmon How would increasing LTchunksize help if the header is in the body?

@FaheemMitha the assumption is that the user did put it at the start, but the header is bigger than the current chunk size. If that's the case, longtable would already have started the output (after the first chunk), and the \end(head|firsthead|foot|lastfood) would've been seen too late.

@Skillmon I see. That's a big assumption.

@FaheemMitha it's not. It's the only sensible pitfall that is not solved after the user saw the first error message. If a user encounters this, what he should first do is make sure that the header is indeed at the start of the table. If he finds it is, the additional help is what will fix the issue at hand.

@Skillmon Shouldn't the message mention that the \endhead (or \endfoot) should be at the start of the table?

@FaheemMitha "Longtable head or foot not at start of table" is exactly that, isn't it?

@Skillmon Oh, right. I'd rephrase with "Longtable header or footer must be at the start of table". But point taken.

@FaheemMitha but that's longer, and longtable comes from a time in which every token was expensive :)

@Skillmon I didn't realise. :-)

@FaheemMitha many of the big packages in use today were unthinkable back when LaTeX gained its popularity. That's a) because we came a long way thanks to ingenious programmers implementing mind breaking stuff, and b) because of the technical limitations of its time. Just consider how many registers TeX had before e-TeX. And think of the memory constraints of the time back then. It is really incredible what people like @DavidCarlisle (and of course many others) implemented with such limitations.

(sorry, the message got too long, so only mentioning @DavidCarlisle since we were talking about the longtable-man)

@samcarter_looks_forward_TUG'21 -- Yes, I'm willing to add that as long as no one has a compelling reason not to. (I might also bring up the great backslash disaster of 2017, which was unintentional but a particularly acute example of bad maintenance practice.)

@TeXnician -- I haven't looked at either the question or the longtable documentation. But from what's been described here, it seems to me that the documentation could be clearer. I'm with @FaheemMitha here.

@barbarabeeton to be fair, @DavidCarlisle's documentation tends to be short in all of his packages. You can't force him to make an exception just for longtable :)

@barbarabeeton Nobody argues about could be. There is little documentation out there that couldn't be clearer. But there's always the question whether the effort will pay off, especially for packages where it seemingly did go unnoticed for decades ;)

@barbarabeeton that would ruin the tradition of leaving David's packages be. :)

@Skillmon they may be short, but with exquisite colour schemes :)

@TeXnician -- A bug was just reported for amsart that wasn't on the existing bugs list (although I'd seen it and fixed it for one book, but apparently didn't properly document it; oops), and that's decades old. The purpose of documentation is to save one's own skin, if only by saying "we told you so."

wait what twitter.com/shouldhaveaduck/status/1420756517838462977

@PauloCereda I wonder why this sign is necessary!

@samcarter_looks_forward_TUG'21 :)

@barbarabeeton Should I attempt a documentation patch?

And if so, would you be willing to offer feedback?

@FaheemMitha -- I'd rather badger @DavidCarlisle. But suggesting text is a good idea, and yes, I'm willing to offer feedback.

@barbarabeeton OK. Thank you very much.

@samcarter_looks_forward_TUG'21 -- Objection posted, at meta.stackexchange.com/a/368003/169475

@barbarabeeton upvoted :)

@samcarter_looks_forward_TUG'21 -- It's gathering support. Has climbed above the bottom third.

@barbarabeeton upvoted too... let's see!

I found a bug. Tomorrow is my last day here, then I'm on vacation... I do not know if doing another mini-release or just wait until a few stuff pile up...

@Skillmon that's not fair. The documentation-to-code ratio of indentfirst is exemplary.

@FaheemMitha \endhead and \endfoot are by definition at the start of the table. The documentation has been largely unchanged for nearly thirty years, some questions about longtable come up quite often but basically no one seems to have problems with the head and foot syntax.

@FaheemMitha what is a big assumption? The fact that chunksize is bigger than the table head? That isn't an assumption it is a documented requirement

ooh directions

@barbarabeeton sort by time and it will nearly at the top :)

@DavidCarlisle documentation is important, see e.g, how easy is to upgrade a PS5 SSD :) ^^

@samcarter_looks_forward_TUG'21 -- While that's true, all it means is that not all that many people are aware of the presence of the question. I wouldn't have noticed except for your pointing it out.

@barbarabeeton The question is not even featured, so little chance to see it

@samcarter_looks_forward_TUG'21 -- The first comment on the question is that it should be featured. I gave that a check mark, but am not optimistic.

@Maïeul In case you are still looking for a contour version of the ducks:

\documentclass{standalone}
\usepackage{tikzducks,tikzlings}

\begin{document}

\begin{tikzpicture}

  \path (0.1,0.08) rectangle (2.12,2.14);

  \draw[overlay,cm={{0.75,0.0,0.0,-0.75,(0.0,2.2)}}] (1.1800,0.0899) .. controls (0.8116,0.0899) and (0.5129,0.4633) .. (0.5129,0.9238) .. controls (0.5129,1.1027) and (0.5584,1.2680) .. (0.6351,1.4038) .. controls (0.3064,1.4888) and (-0.1718,2.0636) .. (0.3648,2.5769) .. controls (0.9185,3.1052) and (3.7990,2.7516) .. (2.4048,1.0319) .. controls (2.1085,1.1911) and (1.9129,1.2733) .. (1.7701,1.3123) .. controls (1.8191,1.1963) and (1.8471,1.0642) .…