I think I found a solution meanwhile; simply \usepackage[urldate=long]{biblatex}.

@psie At a guess, because the format was written by someone who used American-style date formatting (which is fine if you write the date in a long format with the month spelled out - "September 10, 2024", but ambiguous if you just use numbers - "09/10/24" - is it 9th of October, or September 10th?). You can look at tex.stackexchange.com/q/376782 for other suggestions on how to format, as well.

@psie do you load babel or polyglossia?

github.com/latex3/latex2e/issues/1460

@JosephWright \mbox{a\addvspace{1in}b}

@DavidCarlisle :)

@DavidCarlisle In other news, formats are now smaller

@JosephWright what should I do with the logo on my website when the noitemerr is no longer a mysterious error??

@mickep Have you ever tried $\mathunderbar{e}_1$ (or whatever is the ConTeXt syntax for this)?

@UlrikeFischer and the million answers here that look like:

@egreg Yes, this is what one gets:

@DavidCarlisle Oh, that is indeed a confusing error :)

@DavidCarlisle In the picture of your underaccent answer, are the bars scaled or something? They look very tiny.

@mickep So something wrong happens with unicode-math

@egreg Could in principle also be something added to luametatex.

@mickep What I get is that the underbar is placed taking into account the whole depth of the atom, including subscripts. I guess you saw the picture

@mickep The “bad” line is obtained by \mathunderbar{e}_1

@egreg Yes, it very much looks like that, and that is indeed not expected. Is that the same in xelatex and lualatex?

@mickep Yes.

@egreg That points probably to unicode-math, then, indeed.

@mickep for some reason the exisiting accent package code forces scriptscriptstyle, I just made it work as (I think) intended, but it is small...

@mickep Yes, I'm going to report the issue.

@DavidCarlisle Always forcing scriptscript style sounds a bit strange, indeed.

@mickep Can you show plain TeX code that produces the right thing? I'm ConTeXt impaired, you know…

@egreg I am plain impaired... But you can look in math-acc.mklx how wideunderbar is defined.

(Maybe something like \Umathaccent bottom \mathovercode\zerocount "0332{e}_1 is called)

@mickep :-D

@mickep I see nothing strange: \Umathaccent bottom 7 \symoperators "00332

@UlrikeFischer blame you for X = \mathcal{H} ?

@DavidCarlisle Don't worry, we'll impement Typst and get rich!

@DavidCarlisle who else? But now that we have a parallel luamml-file, shouldn't it be possible to catch such mismatch by comparing hash and source?

@UlrikeFischer well comparing even with the existing dummy (now I have re-written it) shows the hash have been shifted. trying to see how...

@DavidCarlisle what do you mean by shifted? That the hash for mml1 is now in mml2?

@UlrikeFischer yes but the first few are right but then they drift off

@DavidCarlisle which file was it?

@UlrikeFischer /arxiv-3/2401.09965v1-tagged.tex

compare the hash with the luamml version (which is checked in) or the dummy you can right by uncommenting the tagpdf setting

eg \mml 15 is $R$ has CCC9B8400A1A5E5030D7C7D09FA64AD1 but the dummy file has that hash on \mml15 but with tex $|\zeta \rangle $

@DavidCarlisle they are all wrong. It looks as if the source is shifted by one. They seems to pick up the following source. Why don't I get a luamml file if I compile. Do we suppress that there?

@DavidCarlisle ah, no luamml because there is no unicode-math. I should read my docu more often.

@egreg And I assume that does not look good in lualatex?

@mickep ooh, German. Wunderbar.

@Skillmon Indeed, and underbar is Swedish for German Wunderbar...

@DavidCarlisle I'm still trying to decide if the old or the new version is wrong. The arxiv-2 looks ok. So perhaps in the old arxiv-3 the starting $^1$ confused things?

@mickep that's wunderbar indeed!

Today's random conclusion: It is not always better to do an extra first run with tolerance 50 instead of going tolerance 100 directly. In fact, it might be worse!

@mickep I produced the picture with LuaLaTeX.

@UlrikeFischer I fixed arxiv3 and updated that and the luamml version

@UlrikeFischer there is a ....-luamml.tex source file

@DavidCarlisle yes, after starring long enough on its code I realized that it loads unicode-math ...

@UlrikeFischer I read the documentation

@UlrikeFischer I'm tempted to upload all the luamml ones to g-drive even if they are not really announced, they may as well be there?

@DavidCarlisle as long as they are "real" (manual) mathml-files the luamml are ignored anyway. So it shouldn't change something, (only the compilation produces a file more)?

@UlrikeFischer no I mean the final pdf with luamml generated AF (there are separately named ...-luamml.tex files already checked in, so that the files have separate names to avoid that issue

@UlrikeFischer so it's just a matter of copyy the ...-luamml.pdf files to g-drive

@DavidCarlisle ah right. Yes, why not uploading them? Perhaps someone checks them for us and find the errors ;-)

@UlrikeFischer done

@DavidCarlisle currently they are then simply hidden in the drive? Btw: did you see the text-in-math discussion?

@UlrikeFischer they are in the gdrive publically visible but not individually linked from the discussion page

@UlrikeFischer you should see them here drive.google.com/drive/u/2/folders/…

@UlrikeFischer text-in-math?

@DavidCarlisle github.com/pdf-association/pdf-issues/issues/467

@UlrikeFischer commented...

@DavidCarlisle very clear comment!

@UlrikeFischer you are not listed as watching that repo (just Chris and Marcel) how did you see that issue?

@DavidCarlisle I have a custom watch (only for issues) on this repo.

@UlrikeFischer oh that's not enough to list you as a watcher, apparently:-) I think I'll use an even more customised watch "wait until @UlrikeFischer pings me"

@DavidCarlisle ;-). Some of the issues are rather curious, but I learned the word "pertinent"

@UlrikeFischer oddly enough I knew that word was pertinent even without watching the repo

can @UlrikeFischer have birthdays more often, please? yesterday was the first time the sun shone for weeks, but now we are back to nothing but rain.

how do I figure out which thing I need to make a comment in which part of a dtx? right now, I basically just guess.

@cfr ?

@DavidCarlisle I think I can't figure out when ^^A is a comment and when it will end up in the files or typeset. or how % interacts with ^^A. (I'm not actually sure this is what I need to ask though.)

@cfr if % is being ignored (so the comments tyoeset then ^^A is a comment) if % is being treated as a comment then ^^A is not

@cfr if you want something to really be a comment just use %^^A and you're on the safe side :)

@Skillmon that was what I thought, but it doesn't seem to work everywhere.

@DavidCarlisle thanks.

but I think it was nothing actually to do with comments and I screwed something up searlier.

@Skillmon or delete the line and write wonderfully clear transparent code that requires no comments

@DavidCarlisle well, that's my usual approach as well :)

@UlrikeFischer Yeah!

@samcarter ;-)