@DavidCarlisle Even cleverer code would tell the user that's what it had just done ;).

@DavidCarlisle So the \??? is just to ensure invalidity?

@cfr yes although you may note my edited comment above my recollection was that it used undefined command (and the doc link I gave says same) but you are getting the doesn't match case. I could check when it used that form, but I'm going to bed:-)

@DavidCarlisle Nos da ;).

\documentclass{article}

\ExplSyntaxOn

\typeout{% expansion context can't use normal errors here
aaa
\msg_expandable_error:nn{wibble}{don't ~do ~this}
bbb
}
\ExplSyntaxOff

\begin{document}

\end{document}

@JosephWright has the above changed? The doc in interface3 implies the underlying error will be undefined command, but it's "doesn't match definition" ? (@cfr) ^^^

ah it changed at github.com/latex3/latex3/issues/931 (we should change the doc)

@DavidCarlisle yes it was better for the number of chats we can have

@JosephWright so we should update:

Hello! hyperref. .out-files have these \BOOKMARK-entries. Does hyperref provide means for hooking into processing them? I'd like to examine the names of the destinations/targets and have some bookmarks not link to a target in the pdf but link to another pdf where target/anchor of same name exists. I thought about patching \ReadBookmarks so that \ReadBookmarks temporarily redefines \@@BOOKMARK. But I wonder if this is really needed.

@UlrichDiez if you want to create other actions, use the bookmark package.

@UlrichDiez the contentsline entries do not create the bookmarks, that is done by the \addcontentsline command. If you want that a \contentsline also (or instead of the normal link) creates a bookmark you should redefine these two commands (and use the bookmark package to actually create them):

\protected\def\Hy@toclinkstart{\hyper@linkstart{link}{\Hy@tocdestname}}
\protected\def\Hy@toclinkend{\hyper@linkend}

@UlrichDiez or instead of redefining the hyperref commands you could hook into contentsline with the new generic cmd hooks with arguments.

@UlrikeFischer The problem is: The same .tex-file is used for creating different .pdf-output-files: a) The entire document. In that scenario before tableofcontents a checkpoint is written to .aux-file. b) a .pdf-file only holding the table of contents. Here\nofiles is used, the checkpoint is read and only \tableofcontents is issued. I managed hyperlinks in the text of the .toc-pdf to be url-links to the pdf-file with the complete document. I did not yet manage this with the bookmarks.

@DavidCarlisle It confused me because the error which followed didn't seem to relate to te \??? error. The 'doesn't match' suggested a control sequence, but the error was an undefined variable in every case. But I assume that's just luck and it could be a cs in other cases ....

@cfr you have to generate some primitive text error we experimented with using an undefined token and other things and finally this does not match error. the idea is to use the least intrusive error possible (as it's unrelated to the actual error), and then put the actual required message somewhere tex will add it to the error context. The options here are extremely limited (unless you use luatex) the main thing is it's not my fault blame Knuth, or Bruno or @JosephWright

Is this a bug in microtype or is the additional spacing expected?

@cfr I started to look last night and then gave up. It's probably the underlying pdftex feature rather than microtype, but I haven't checked.

@DavidCarlisle Thanks, yes. That's why I said it was probably luck ....

@DavidCarlisle Would that affect luatex? It's not pdftex-specific.

@cfr possibly (the code all originates from pdftex originally) but as I say I gave up, may look later

@DavidCarlisle Ah, I didn't realise that. (And it can't be tested with a third engine because the feature isn't supported.) Thanks.

@UlrichDiez if you know which document you just produce, it is trivial to enable or disable the needed code for the bookmarks. You can simply read in the toc, extract the data and create the bookmarks.

@DavidCarlisle Bruno

@Skillmon e-type plan is looking good I think: expect a branch to appear during the next few days (I've done all of core expl3 except l3basics, so that plus the docs is a minimum)

@JosephWright skype:-)

Interface 3 p32: 'For backward compatibility it is currently possible to make ... or N or c-type variants of an n-type argument. Both are deprecated. ... The second because programmers who use that most often want to access the value of a variable given its name, hence should use a V-type or v-type variant instead of c-type. In those cases, using the lower-level \exp_args:No or \exp_args:Nc functions explicitly is preferred to defining confusing variants.'

What does 'those cases' refer to? It seems to refer to cases where people want to access the value of a variable, but that would suggest avoiding V/v, which seems at odds with the previous sentence's endorsement (and the next page and elsewhere). Have I misunderstood or is the reference to 'those cases' to something else?

@CarLaTeX Your city gets mentioned tex.stackexchange.com/q/696742/36296 :)

@samcarter The Politecnico of Milan is the best university in Italy: polimi.it/apertura/dettaglio-apertura/home/…

@samcarter and my cousin's ex-boyfriend is a professor there :D

STOP EVERYTHING NOW

@JosephWright ^^

@CarLaTeX The better the university, the worse the templates? :) (it is actually not that bad, just some weird logic about frame numbers)

@PauloCereda ohh

@samcarter ooh

@samcarter Imagine the others :D

@samcarter I think those "bars" were a bit thin, also. Moire vibes...

@mickep I thought they looked like the two sides of a zipper :)

@samcarter Oh, that is a more fun interpretation.