@DavidCarlisle, @UlrikeFischer Running final checks on the dev -> main merge locally here

Hello, I put the word "example" in the table by the following code \newtheorem*{example}{\begin{tabular}{r|}example\\\hline\end{tabular}} How can I do that for sections title? Could anyone help me, please?

But it doesn't work properly

@JosephWright: \inteval seems to be part of the kernel. Where can I find its documentation, apart from texdoc xfp? (Out of curiosity.)

@user91500 You should ask a proper question including a MWE on the main site. How to change the layout of section titles depends a lot on the class you use.

@AlexG It's in usrguide

@JosephWright Thanks a lot! Nevertheless, documentation of kernel built-ins seems to be spread about multiple manuals, e. g. the ltfilehook commands. Is there a summary of all packages that are part of the format?

@AlexG No, because they are aimed at different people: usrguide is for end users, clsguide for package/class authors; we aim over time to add mainly to those two places (they've just been revised heavily)

@JosephWright Ok, I see. Thanks again!

@AlexG Hooks are I think the only thing that's not in either of those: perhaps one to ask Frank about

What programs are to use to create such font symbols?

Any other Tom Lehrer fans out there? See tex.stackexchange.com/a/688045

@barbarabeeton OH! <-- @PauloCereda

In the table of contents, I want to remove the dotted line that appears between the toc entry and the page number. I can do this with \renewcommand\@dotsep{1000}, but is there a more elegant way for that?

@yo' -- Yeah. It includes some that he never recorded, and that I don't know. Now ... to find the music ...

@tush -- Depends on what document class you're using.

@barbarabeeton Thanks. book

@tush define \l@section so it is not \newcommand*\l@section{\@dottedtocline{1}{1.5em}{2.3em}}

@DavidCarlisle Thanks. Can you give me a suggestion? I actually want to use those two lengths that @dottedtocline allows me to insert. I use them to vertically align the sections and subsection nicely one to the other.

Against my better judgement, I've posted an entire answer using expl3. Criticism welcome.

I wonder why Knuth chose f/1000 for stretching and 1000/f for shrinking (spacefactor). One can read this on p. 75-76 in the TeXbook, but I see no argument there on why one should in general shrink much less than one should stretch (for f large). Anyone know if he elaborated on that somewhere?

@AlanMunn Not understanding it, but the outcome is fine, so +1. :)

@mickep Thanks. :)

@mickep -- Possibly covered in his joint paper with Michael Plass. (Just a guess; it's been a long time since I read it.)

@barbarabeeton I am waiting for Digital Typography to arrive, but it seems to be stuck in the mail somewhere.

(Thanks!)

@AlanMunn Variables should be in a module, so \l_<name>_authformat_tl, etc.

@AlanMunn Things like \cs_new:Nn \l_map_author_info:nn should use \cs_new_protected:Nn as the content is not expandable (there is typesetting in it)

@JosephWright Ah, right. Thanks fixing now.

@AlanMunn At the 2e level: I'd have the \par inside the {...} pair in case the formatting changes the baselines

@AlanMunn We have \seq_set_from_clist:Nn, you don't need the use of a temp. var

@JosephWright I must have missed that version. That makes things simpler. Is there any reason why there isn't a \seq_get_item:NnN to put a specific sequence item into a token list? Seems like a natural thing to have.

@AlanMunn Log an issue ;) (Mainly we use sequences 'in the whole' for mapping and similar, so this has not come up)

@JosephWright Ok. I can do that. Thanks a lot for the comments.

@cis Maybe a bit late, but you can use the developer tools, select the element and then select the "Computed" tab. At the very bottom, you will find the "Rendered Fonts" which are those used for the relevant text node.

In my case it is "Segoe UI Symbol"

@JosephWright isn't that unfair. You were just nearing 100 ...

@UlrikeFischer :)

@UlrikeFischer Don't worry, we'll get there!

@cis But actually, the very latest version of DejaVu Sans (available here: dejavu-fonts.github.io) should also have these glyphs. But maybe there is only a version with limited glyph support installed on your machine.

(see: dejavu.sourceforge.net/samples/DejaVuSans.pdf)

@JasperHabicht @JasperHabicht "Deja Vu Sans" works. But it's glyph are not so spectacular and they are not the shown one at wikipedia.

@cis That's true =)

@barbarabeeton Not having the skills to give a relevant criticism of your answer, I can simply indicate that among the 40th posts that I consulted on the use of %, I do not remember having seen anything concerning the addition of spurious space by a package like in you e.g... So your answer seems to really fill in a blank IMO!

@barbarabeeton Tried now with f/1000 in both places, and sometimes the linebreaks get better, but also the contrast of the space after the period on consecutive lines might be too large.

@SylvainRigal -- Thanks for your feedback. I happened on a question with an example like that, that was nominated for deletion as a duplicate of the % question. It clearly wasn't a duplicate, and it had a decent answer, so I challenged it in a comment. After some discussion, the attempt to delete as a duplicate was removed, but so were the comments, and apparently the question a well. But it's an important exception, so I created that answer to point that out,

@mickep -- Ah, the "rivers and puddles" problem. (I think most typesetters refer to these as "ponds", but I like "puddles" because those are more random.) I think this was addressed in a TUGboat article, or maybe at a conference talk. I'll look later; I'm trying to write my talk for this year's meeting. (Rivers aren't addressed by K&P.)

@barbarabeeton Hmm, but aren't rivers the "gaps" on several consecutive lines?

@mickep -- Yes, but that's how I read what you're describing. Rivers and puddles are related, so I think of them together.

In this case, the more possible shrinking (to the right), makes it possible to move the "your" up one line, and therefore the overfull hbox is avoided.

(The left one is the "standards" one)

(Good luck with the talk preparations! I'm off reading for the kids.)

@mickep -- I agree. I like the one on the right better.

Building LaTeX2e 2023-06-01

@JosephWright Very good!

@mickep I had a few bits of 'fun', I think I'll adjust the recipe a bit for the next time

@JosephWright Oh, why avoiding the fun? :)

(I'm not sure I understand what it means to build LaTeX2e 2023-06-01)

@mickep Extract the code, run all the tests, typeset all the docs, create zip files, make available on GitHub

@mickep Nowadays, it means I push a tag to GitHub, which triggers the automatic process

@JosephWright Ah, and when will that be available in texlive?

@JosephWright So, pretty convenient, then.

@mickep I'm a bit late in the day today, so CTAN probably won't install until tomorrow, so TL not before Saturday

@mickep Yes, we've worked hard to automate this

@mickep github.com/latex3/latex2e/actions/runs/5215037425/jobs/…

@JosephWright 26 minutes and counting, so I guess that is the big run?

@JosephWright -- But it's already 2023-06-08.

@barbarabeeton That is correct.

@mickep Ues takes about an hour nowadays

@barbarabeeton :)

@barbarabeeton Nominal date - vote or more as a version string ;)

@barbarabeeton We had a few things that needed the extra days

@JosephWright -- Oh, I'm not complaining, just teasing. I assume the "nominal date" will be entered as a literal.

@mickep it has some logic as long as you think of large f as allowing large space like after a sentence, not thinking of it as allowing large flexibility

@barbarabeeton We do aim for the date: we did talk about changing to mid-month, but it’s non-trivial adjusting the sources

@DavidCarlisle So, merely something to allow large space more than to increasing the solution space. I still do not really get it, but doing some testing now. One could also think of f/1000 and 1 (instead of 1000/f).

@mickep would using an inverted formula provided some mathematical amusement for the software author be a better answer?

@DavidCarlisle Might be closer to the truth. :))

ctan.org/incoming

@JosephWright rollback test?

@DavidCarlisle Yes ...

@JosephWright that doesn't block the build?

@DavidCarlisle I sorted it

@JosephWright ah I guess commit mails arriving out of order as I got the failure notice just now:-) (rollback tests are ... interesting)

@DavidCarlisle Rollback was wrong on dev, but I sorted main a while ago - I'll try a different order of steps next release

@DavidCarlisle Both branches now working, tags pushed, building or built for CTAN ...

@JosephWright why did you have to update so many tagging tests? They all worked in develop.

@UlrikeFischer That's why I will try a different flow next time

@UlrikeFischer Current method is merge first, then change the version string. Issue arose as that means anything that tests the kernel version then changes what it does (works on dev, then doesn't work on main, then works again on main).

@JosephWright ah right, there is a ifformatlater test in the testphase file.

@UlrikeFischer Yes - previously we've not had that (or tested for functionality not format version), so it's not come up before

@JosephWright the problem was that the phase-III file is new, so even if installed in dev it is also found by older latex and then errors as the new code uses stuff that doesn't work, so I thought adding a check would be safer. Testing for e.g. some hook with arguments would probably have worked too, but the format version looked more logical.

@UlrikeFischer Sure: I just need to allow for that in the workflow next time