@JosephWright -- Interesting point of view. But I would offer a different reason for replacing ...\over... with \frac{...}{...}, namely, scoping is much easier for an author when there are nested fractions or complicated numerators or denominators. I do understand Knuth's reasons for preferring \over -- I asked him personally and his answer was clear -- but having been faced with input from authors using both plain TeX and LaTeX, it's obvious that there are fewer glitches with \frac.

@barbarabeeton Yes, I think the only nice thing with \over is that it reflects the way we read math. (Is there any other macro that has that strange format?) The reason to not support \over was thus not only about spacing, but also structure/grouping/mathstyle. The braces { } are by the way now grouping (and not enforcing an ord class).

@samcarter oh no twitter.com/shouldhaveaduck/status/1544859658636369920

@PauloCereda oh, a mighty duck :)

@samcarter ooh

https://texlive.net/run?\MakeTitlecase{blub} @JosephWright @UlrikeFischer

@DavidCarlisle World Blub Domination!

@DavidCarlisle @JosephWright do we actually expect \@ifpackagewith to work with keyval? (currently it doesn't if the key has a value). And if not, what can one use to test if a certain option has been passed?

@UlrikeFischer No, I don't think we do

@UlrikeFischer it's a bit hard to define what it should mean. For the unused option list we just store the key name, which works well enough as it's only a warning, but if you say draft=false which is functionally the same as not using the option, what should \@ifpackagewith report?

@UlrikeFischer It just doesn't work the same with a keyval-based package: I think people simply have to pass the keys and leave any clash management to the package

@JosephWright if you view legacy non kv options always as boolean keys, then \@ifpackagewith is essentially \ifbooleankey{..} and only usable on boolean keys (and tests true however that key set, not just as an option)

@DavidCarlisle yes it doesn't really work. I was just pondering over hyperrefs \@ifpackagewith{hyperref}{tex4ht} and wondering what to do with it. It would probably still work, as driver options have no value, but the code seems only to be there to add the tex4ht driver if someone forgot it.

@UlrikeFischer i'd use a \if@blub set (or not) by the option, it will be clearer that way

@DavidCarlisle the options haven't been processed yet. hyperref is testing if it should extend the list. But on the whole I think it is the job of the tex4ht scripts etc to pass the correct option to hyperref.

@UlrikeFischer yes or simply drop the outer test: if tex4ht package loaded set the tex4ht key true. Then if the options have tex4ht=false and it gets set back, so be it

@UlrikeFischer ooh

@DavidCarlisle tex4ht.sty contains \expandafter\def\csname opt@hyperref.sty\endcsname{tex4ht} and the option is set. So the test for the package is unneeded (unless the order is relevant), and it remains the \ltx@IfUndefined{HCode} test.

@UlrikeFischer OK

@UlrikeFischer yes supporting specific drivers is one thing but patching other packages...

@samcarter ooh Matrix secrets

@PauloCereda oh, secrets :)

I am having difficulties using cmbright.sty with lualatex (TL22). While I get the expected output with pdflatex, lualatex replaces the font with lmroman and I get LaTeX Font Warning: Some font shapes were not available, defaults substituted. Is there something misconfigured or is cmbright obsolete?

\documentclass[border=1pt]{standalone}
\usepackage{cmbright}
\begin{document} Hello! \end{document}

@AlexG it doesn't support TU encoding.

@PauloCereda I killed 54.

@Skillmon ooh

@Skillmon also, sorry for the delay in replying to you, the last few days are very hectic... will do that today, and I have good news. :)

@Skillmon 60. I'm meaner

@UlrikeFischer I think it worked a while back. Has it changed in the meantime? Can I fix this somehow?

@AlexG nothing has changed here for a long time. Perhaps you used it with T1 or OT1 encoding, but this is not recommended for lualatex.

@UlrikeFischer If I add \usepackage[T1]{fontenc} the PDF output is blank when lualatex is used. With TL21 I get at least something that looks ok (sans font used), though the fonts listed in the PDF viewer's document properties dialogue are different for pdflatex (HFBR10) and lualatex (SFBMR10).

@AlexG T1 and OT1 use different fonts (as they do for cmr), that is normal. I see no difference between luatex and pdftex. This here works fine with both engines:

\documentclass{article}
\usepackage{cmbright}
\usepackage[T1]{fontenc}
\begin{document}
Hello!
\fontencoding{OT1}\selectfont
Hello

\end{document}

@AlexG I get hello typeset with T1 encoded luatex

@DavidCarlisle So, who will take over now? You? @JosephWright?

@mickep me

@mickep A DUCK

@JosephWright Oh

@UlrikeFischer , @DavidCarlisle, thank you, both. It seems to be a PDF viewer problem. I changed to okular and now the page isn't blank anymore.

@PauloCereda everything's fine, no harm done :)

@JosephWright The usage handler is ignored if one use a family name with a slash in it \ProcessKeyOptions[testnewkeys/blub]. If that is unavoidable as it is viewed as a path, one should probably document that.

@UlrikeFischer oh no! I used that in the past to separate load-time only options from other options :( Now I have to check whether my old classes still work :(

@Skillmon well old classes hopefully don't use the new key val handler ;-)

@PauloCereda ohh twitter.com/scottygb/status/…

@mickep haha

@UlrikeFischer no, they use l3keys2e or \DeclareOption*.

@Skillmon SO no change then

@samcarter ooh trendictoc.com/products/…

@JosephWright yes, but if you change their support is what I meant.

@Skillmon the keys still work. Only the new special .usage:n handler doesn't work as expected.

@UlrikeFischer good :)

@PauloCereda A \Huge duck!

@UlrikeFischer Hi, doubt with hyperref repo, when I run l3build doc I get error...problem from my side?

@Skillmon We've not altered existing behaviours

@PabloGonzálezL not my fault. This here errors with make4ht:

\documentclass{article}
\pagestyle{headings}

\begin{document}
\section{abc}

\end{document}

@UlrikeFischer Ouchh

@UlrikeFischer Thanks, I thought I broke something on my local machine :D

@PabloGonzálezL It breaks at the \MakeUppercase in \sectionmark.

@UlrikeFischer \let \MakeUppercase=\pain :D

@UlrikeFischer lots of newkeyval:-)

@DavidCarlisle ?

@UlrikeFischer lots of mail from github with diffs showing l3keys creeping over hyperref:-)

@DavidCarlisle :)

@DavidCarlisle oh. Sorry ;-) but it has to go in steps, there are tons of keys and I also have to consider all the definitions in the generic driver.

@UlrikeFischer no it's good actually: I have actually read many of them, a single 1000 line diff would have got marked as read...

@mickep -- Understanding braces is a basic requirement for using TeX effectively. Scoping is a "first-order" purpose, but other uses are also important. I'm mulling over writing an introductory essay on "TeX" basics. I'll continue this discussion offline.

@barbarabeeton I look forward to read that. Also, thanks for telling Karl about Lucida.