Quick L3-Question: Does the s argument type also work with \NewDocumentEnvironment or do env and env* have to be defined separately?

@Lupino separately, or at least s would define \begin{env}* whch probably isn't what you want.

@DavidCarlisle that's what I suspected, thanks fpr clarifying

@DavidCarlisle imho, one could add an argument type to ltcmd for that. Something like S, must be the first argument, only applicable for environments. Then transparently define an environment <env>* that has #1 as true and an environment <env> that has #1 as false, and use the same underlying <env> code function.

@Skillmon We did try auto-coercing s for environments - it didn't work that well

@JosephWright maybe I'll take a look (right after I did the tl_act PR which I haven't even started, the etl feature I contemplated, the expkv-cs rewrite I planned, the l3fp support for pgfmath for which I was taking a first look for 5 minutes yesterday, and probably some other stack items I forgot already)

On a time sink thing: is it possible to tell lualatex to continue and build the PDF in case of this error:

Related to this: tex.stackexchange.com/a/737051/38080

@Skillmon :)

Ah, it seems it's a feature. github.com/latex3/luaotfload/issues/119. Any way to include at least one glyph from the font in tex.stackexchange.com/a/737059/38080?

btw, are there plans to upgrade the output routine in the near future? Specificly: allow more than two columns natively?

@Lupino there is a PR currently open that reworks aspects of the output routine, so answer to the first question: Yes. Allow more columns? No, I think not.

@Lupino see github.com/latex3/latex2e/pull/1659

@Lupino just having more columns would probably be easy to add as a package (although I'm not aware of one) the complication would be in how fancy you get in allowed float regions. If you only allow 1-column or full page width floats it's probably a simple extension, if you allow floats that span two out of three columns it gets complicated very quickly as you get combinatorial growth in the number of float placement areas.

@DavidCarlisle well, multicol… but yes, this would be a whole new chapter…

but i also thought of asymmetric columns like here: human-biology-and-public-health.org/index.php/hbph/article/view/…

@Lupino that isn't two column in that sense (text doesn't flow) they are marginpar with a wide margin

bad example, in this paper yu can see it better: human-biology-and-public-health.org/index.php/hbph/article/view/…

keywords should usually be in te left column and the abstract flows into the balanced second page

so far, the editors have to do the breaks manually, but i managed to reduce it to a single macro they have to pass down as PIs.

@Lupino that's still a two column document with equal column widths and a handcrafted first page

@Lupino ah that is changing the text width at a column break. You need alice for that but she's been hiding. @mickep is going to pop up and say it's easy with luametatex's multi-pass paragraph breaker, but in classic tex that's somewhat hard.

@Skillmon this paper is completely automated… all typesetting is done in Word with PIs that are passed down to tex, but the tex code itself is oblivious to the editors.

@DavidCarlisle No, that does not seem reliable.

@DavidCarlisle Paragraph building first, still.

i remember asking a question here but i never got it to work properly for this journal

@Lupino But you do it automatically in the journal? Are you using parshape there?

@mickep no, it is done "by hand" with a single macro that detects if it breaks from left to right unbalanced column or from unbalanced right column into balanced left column. But the macro has to be injected by the editors if they want the break mid-para. otherwise this is done automatically.

@Lupino Ah, that makes more sense, indeed.

@DavidCarlisle Did you break pdftex? tug.org/pipermail/pdftex/2025-February/009443.html

@mickep the SE.anwer seems to suggest that it is possible with lua, but there were some nodes i couldn't figure out were still missing

@mickep well in a way, someone submitted a test on tagging and I ran the test and got other issues (like most of the math was all white) so we decided that wasn't a tagging issue, an @UlrikeFischer cut it down to the change in real space handling, but pdftex is pretty unuable in tl2025 at present.

@mickep Of course for us the luatex change is far more of a pain, the output is correct but the log changing breaks dozens of tests. the pdftex change is no bother at all: the log doesn't change, just the pdf output isn't readable. But who looks at the pdf?

@Lupino I didn't look at it.

@DavidCarlisle Oh, sounds hairy. Is that the same problem that Karl writes about?

Without causing too much visibility - where should I report that nonstopmode should not prompt for anything even in special cases -- code causing some interaction: tex.stackexchange.com/a/737059/9075

@DavidCarlisle Oh, priorities. :)

@koppor use batchmode instead? (I'm not sure I understood the issue)

@DavidCarlisle well it does change lots of pdf-based tests but that's me who has to handle them ;-(

@UlrikeFischer yes there are not so many pdf tests in the core repo, more in taggng I guess

@DavidCarlisle ahhh, revenge :)

I am tempted to come up with an answer of a table consisting of about 15 columns and four rows ... tex.stackexchange.com/questions/737119/create-a-special-table

@JasperHabicht will the rules in your table be as curved as the ones in the image? :P

@samcarter Sure! And the text will also be as blurry! A true challenge!

Oh, some code came along

@samcarter two of your favourite packages combined:

@DavidCarlisle The pinnacle of beauty!

@samcarter you do need this week's ctan release version to get the full effect.

\documentclass{article}
\usepackage{amsmath}
\providecommand\ttrotatenormal{22}

\usepackage{colortbl,typewriter}

\raggedright
\begin{document}

\begin{tabular}{|*4{>{\columncolor{red}}c|>{\columncolor{yellow}}c|}}
  \hline
  one&two&three&four&five&six&seven&eight\\
  \hline
  1&2&3&4&5&6&7&8\\
  \hline
\end{tabular}
\end{document}

@DavidCarlisle ... of course updated immediately as soon as your new version became available :)