@user202729 and also an unsolicited advice from someone who burnt hia fingers before: Do not omit the braces for an n-type arg, even if it happens to have only 1 token.
@user202729 We document two places that it is expected to work: a) with \use:n and so forth; (b) with \int_compare:nNn(TF) and so forth. All other cases can be implemented such that braces are required
@DavidCarlisle picture doesn't trample on @tempdimc, cutwin shouldn't use it like it does. I don't think that picture should use l_tmpa_dim, as the docu imho says that the kernel doesn't use it. But perhaps we have some internal version?
@UlrikeFischer I could take to the team list?, I could use something else \@tempskipa for example that has the same status and just avoids the issue here although my preferred solution would just be to fix cutwin and ask Peter if we can send it to ctan
It looks like that even with expl3, TeX is still not as powerful as a "normal" programming language? For example, result value from a non-expandable function cannot be fed into another function.
On the other hand it can be "fixed" by changing the API of the function to assign the result to some variable, right? (like assembly)
@user202729 all turing complete languages are powerful, just some are less convenient than others. You can inline expandable functions and can inline non expandable ones sometines, so your general statement isn't true
How do I search for the main definition of a function name in interface3.pdf? Sometimes a function is referenced 12 times and iterate through them (...actually using PDF viewer's search feature isn't that bad)
The TeXbook has a underline and a italic (probably one of them is the "main" definition)
@user202729 1. The standard says so. 2. It's consistent. 3. If you (or someone who uses your code) ever change that to nx while copy-pasting, it will look innocently correct but be utterly wrong.
@user202729 looks like we don't distinguish that in the index, but if you know the function you know the prefix so just going in via the table of contents to that section will get you threre as they are in order
That's why you should understand what are you writing when you modify some code, to fix possible bugs that pop up like this / but yes I can see why you should avoid undocumented behavior.
Okay `cs_generate_variant actually work with the wrong number of variables (error checking is slow and requires more work to implement, I guess?)
@user202729 not grabbing arguments is endemic in Tex eg you might document that \section takes an argument for the section head but it is actually a macro with no arguments. If you implemented it as a macro that grabbed all its documented arguments and passed them on to the implementation functions it would take more space and be slower.
@UlrikeFischer I was looking again at colorants: I think one issue is I'll need to re-think how to generate the separations for dvips. I'm going to look this evening
@UlrikeFischer If we need them, I will of course sort it: I think we are pretty close to being able to cover all of the other colour space packages
Just received a swag email. It's weird because it's for achieving 100k rep this year or since March 2019. But I've been above 100k a lot longer than that.
@AlanMunn I got it too. I got also one in august for having 250k reputation (and I got a T-shirt ...). I suspect they are working back to smaller numbers now.
@DavidCarlisle yes Will has imho the largest part of the packages. I think one should correct cutwin here, even if picture is changed: its use of the dimens is wrong, and could e.g. break again if someone uses them in a hook. Why is Frank keen on changing picture?
@UlrikeFischer we stopped using \@tempa macro names when we made 2e and introduced \reserved@a as even tempory uses were tripping over but we couldn't afford to do that for registers but we could with etex and so could have reserved@tima or whatever, but I'm not convinced
@JosephWright Yeah, I'm not sure I'm going to ask for any. The previous T-shirt was pretty low quality, and I'm definitely not going to use the socks. I'd be happy with another coffee mug, but they don't seem to be on offer any more.
@JosephWright I think I have half tired eyes, but I think the sources in the l3backend implementation code documentation (updated as of today) is slanted (LM mono slant), am I right?
@PabloGonzálezL The sources are slanted for all the conditional parts (everything which is surrounded by docstrip guards and therefore only included in some of the generated files). In l3backend, that's pretty much everything since docstrip guards are used to separate the code for the different engines and the backend is of course very backend specific.
@UlrikeFischer I wondered about setting up a new gh organisation but one-file organisation sem silly. I'm still tempted to merge ho-tex and rf-latex and one or two others into a general aquired latex packages github organisation where we could add things like this