Is it reasonable to always be loading expl3 by only calling xparse or would first loading expl3 be better practice (particularly in terms of what will yield the desired results long-term)
@DaiBowen Usual rule with packages is load directly ones where you use their documented features, so if you only use \NewDocumentCommand just xparse but the moment you have \ExplSyntaxOn that means expl3
@DaiBowen In this case, of course, xparse is known to be entirely dependent on expl3 so it's slightly complex: the other example is TiZ/pgf where you can't load the former without the latter
@egreg I've risked going to see La Traviata next Sunday. My cousin's daughter has just broken up with her nth boyfriend and wanted me to accompany her, but she did not manage to get the under 30 discount for my ticket, so this time I escaped. I don't think I'd be able to stay awake for 3 hours of opera!
@PauloCereda She broke up with the first due to his mother, with the second because it was too selfish (I didn't like him, too), with the third because he always wanted to go out till 6 a.m. with friends...
@CarLaTeX: Cricket is easy: you have two sides, one out in the field and one in. Each man that’s in the side that’s in the field goes out and when he’s out comes in and the next man goes in until he’s out. When a man goes out to go in, the men who are out try to get him out, and when he is out he goes in and the next man in goes out and goes in. When they are all out, the side that’s out comes in and the side that’s been in goes out and tries to get those coming in out.
And sometimes there are men still in and not out. There are men called umpires who stay out all the time, and they decide when the men who are in are out. Depending on the weather and the light, the umpires can also send everybody in, no matter whether they’re in or out. When both sides have been in and all the men are out (including those who are not out), then the game is finished.
@CarLaTeX it's more boring and takes 5 days. Go to the Opera instead (it's in some weird language, but it's Ok they put a translation over the stage in most places:-)
@DavidCarlisle Well, the error must show, for the reason I explained in my answer; in the test with \dim_compare:nNnTF, the right hand side is a macro declared with \NewDocumentCommand` that will definitely not work with the latest xparse.
@DavidCarlisle As soon as you have \sidenote the problem shows.
@egreg, @DavidCarlisle Unfortunately it seems there have been quite a number of uses of \NewDocumentCommand in places where this really isn't appropriate
@DavidCarlisle Package versions change only when I do a release, so if you are using the SVN you won't see it from \listfile but will from kpsewhich (this is a reason I've been meaning to write an extended \listfiles using MD5 sums)
@egreg I see this with the current release but not the burning edge: that's as-expected (as @DavidCarlisle notes)
@JosephWright yes still seems a bit dangerous to have different code with same version number. (or course in the 2e code we always bump the version by hand on edit so it doesn't arise there)
@DavidCarlisle @egreg How would the \DeclareTextCompositeCommand on the list look if one had etex and expl3? (I'm planing a small task about expl3 and this looks like a good example ...).
@UlrikeFischer not sure:-) also my detokenize suggestion doesn't work, it has to be \@empty as that is what is used in the actual command (it;s even documented in ltoutenc, I found after trying detokenize)
@JosephWright build install in base just put me in a tight "yes" loop :(
(Press Enter to retry, or Control-D to exit)
Please type another input file name: ! I can't find file `y'.
l.39 \input ltdirchk.dtx
(Press Enter to retry, or Control-D to exit)
Please type another input file name: ! I can't find file `y'.
l.39 \input ltdirchk.dtx
(Press Enter to retry, or Control-D to exit)
Please type another input file name: ! I can't find file `y'.
l.39 \input ltdirchk.dtx
@egreg As @DavidCarlisle says, no but we've avoided having a \let hidden inside the set up for those cases. The problem is these uses are not supported.
@Johannes_B At one point we had all of them renamed internally 'again' so they really weren't used, but that gets somewhat tricky (at the low level you do need to know the primitive behaviours)
@CarLaTeX -- don't listen only to @egreg -- his appreciation of music ends with the death of beethoven. your experience may vary, but there's lots of more recent music that is worth listening to and enjoying (even if it takes some effort the first few times you hear a piece).
Are there any common reasons why synctex just wouldn't sync to the correct position from syncing from PDF to .tex file? It's a couple of paragraphs off.
I'm using TeXShop. I use microtype, fourier, geometry, hyperref, inputenc packages.
I won't have time to try to reduce this to a minimal example today, but if there is some extremely common cause that I didn't find, I'd appreciate a link
@Johannes_B Very funny! Unfortunately, my English is not good enough to understand everything, but just seeing an English TV show about cooking it's comical... OMG, I've put cooking and English in the same phrase!
@DavidCarlisle Surely he'd enjoy! It's full of beautiful models!
would it work simply to use l3build to bump the date after you've done a ctan build (just as in 2e I always bump the patch level, so I can tell if I'm using a format build from svn)
@JosephWright oh I suppose that's option 3 in your list...
@DavidCarlisle A few more possibles came to me whilst out for a walk with my nephew, but probably they don't need to be raised immediately: they are all relatively obvious
@DavidCarlisle Yes, can simply go for a policy change
@DavidCarlisle You'll notice the lack of any comment on the version control system used :)
@DavidCarlisle Probably I should tag the releases anyway
@UlrikeFischer that's a bit odd I suppose I should look inside the box and see what's different. Not just now though.
@JosephWright given the dates are all normalised before a release presumably it doesn't harm if making a change to change the version anyway, it's what I've always done editing dtx files so not doing that would seem odd if I got back to editing in l3 area
@DavidCarlisle Certainly another possible: that of course overlaps with the question about how one expects sources and 'use' code to related (cf. Git and the fact the lack of $Id support is by-design)
@JosephWright another option might be to instead do the replace in the build area not the source so that if you just do texlua build.lua install you get whatever the versions in svn say they are` but if you do texlua build.lua ctan --date 2017/02/19 it builds a verson with normalised dates. perhaps....
@mirabilos if your unicode commands were correctly defined as robust commands they would work in latex write commands such as \addtocontents (but not in a primitive write)
@DavidCarlisle yeah, but I mean, ... oh I don't know what I mean. I think it's that you knew well what you were doing :-) (does that hide under moreorless? I hope so.)
@yo' It's generally impossible to emulate a primitive via macros; the syntax for \write is \write<number><general text> and <general text> is <filler>{<balanced text><right brace>. So in theory you can say \write20\relax\space\relax\bgroup whatever}. With macros it's quite difficult (to say the least) achieve such syntax.
@mirabilos delcarerobust perhaps, depending what you do, but in fact it's really quite hard to write the expansion of a command defined by \protected\def to a file.
@mirabilos too many disconnected fragments:-) If you post a MWE I might look (or I might eat instead, depending:-) are you just padding with 0 for cases where i didn't bother to do that in the standard input enc error message display of the hex?
@mirabilos if you want that may as well fix the version in inputenc to do it, I always find it mildly annoying when it shows three digits U+F56 or whatever, i just didn't do it at the time to save a few tokens in what is only an error case anyway
@JosephWright actually I did add a couple of \protected to the file but it's hacking in to inputenc/utf8 and by the time you realise it's not safe you don't just want to suppress expansion but rather do something completely different so without re-writing it all to fork earlier that works as well as anything
Package inputenc Warning: Unicode char 💩 (U-0001F4A9) not set up for use
(inputenc) with LaTeX, replacing on input line 20.
Package inputenc Warning: Unicode char ই (U+0987) not set up for use
(inputenc) with LaTeX, replacing on input line 22.
Package inputenc Warning: Unicode char ⮢ (U+2BA2) not set up for use
(inputenc) with LaTeX, replacing on input line 22.
I have a paragraph of lots of vectors symbols, so I need to type something like $\mathbf{a}\cdot\mathbf{b}=\frac{\mathbf{a}+\mathbf{b}}{\mathbf{a}\cdot\mathbf{v}}$(just an example). It is so inconvinient to write a \mathbf command everywhere. So I want to define a command that automatically distr...
@JosephWright hm okay, there’s probably not too much to document though (unlike for the other package, which can ship a list of supported Unicode codepoints)
how do I include the PDF? latex source plus precompiled PDF?
mathtools amssymb ascii bbding esint extraipa ifsym marvosym mathdots mathrsfs pifont pmboxdraw textcomp tipa tone wasysym xfrac, and a subset of stmaryrd
this is my “let's declare as many unicode characters as we can" package
(and we don't even math mode in any documents… except of course in the expansion of the Unicode codepoints, but the mere loading of the package seems to trigger it)
@mirabilos well if you used a unicode tex you'd only need one family (as the font can have 10FFFF characters, but you can have 256 fonts of that size), but you said you wanted to stick to pdftex so you get 16 times 256 instead
@mirabilos math alphabets like \mathscr etc are dynamically allocated just if used but symbol fonts have to be allocated at definition time as \mathchardef\foo"1234345 requires a fixed bit-packing of the font number and character number
@mirabilos what are your documents, if you are not doing much math you probably don't need many or any math alphabets at all. just declare the fonts as text and then if you need a rightarrow use \text{→} then if you have set up → to work in text mode it will work in text and math using the text fonts (there is some issues of math spacing but...)
@yo' thing is many of them were wrong so if he has to make a tds layout and then check the supplied one, it's easier for him just to make a tds layout, but with l3build of course that could never happen...
@JosephWright test failure:
! ^^I./ltluatex.lua:357: in function 'create_callback'
shoudl I update tlb-ltluatex-001.luatex.tlg or shoudl we normalise away those line numbers in l3build?
@JosephWright only from files, even in that test there is a bit from directlua
stack traceback:
^^I[C]: in function 'error'
^^I./ltluatex.lua:109: in function 'module_error'
^^I./ltluatex.lua:116: in function 'luatexbase_error'
^^I./ltluatex.lua:353: in function 'create_callback'
^^I[\directlua]:1: in main chunk.
@JosephWright I think we're eating in a couple of minutes so I'll check later if you are around
since I now have two packages, which can be used independent of each other, should I upload them to CTAN as two separate packages? (If not, I could use unicodedomino in unicodepoints’ documentation PDF)
@mirabilos I think the "do something more interesting with undefined slots" package is probably more generally applicable than a "load every font package there is" package. so if that's the split you mean, I'd split.
@mirabilos that's a possibility you can put them together on ctan but people generally get confused by "package" in the sense of ctan and system "package managers" which means "bunch of files installed as a unit" and "package" in the sense of latex \usepackage which means "single file with extension .sty` so probably I'd upload then separately although I have several ctan "packages" that consist of multiple files so it's up to you....
I sort of got to go (drink cocktails to celebrate… also it’s already 21:15 here), but I already published the .sty files themselves on github.com/mirabilos/tex-unicodedomino and github.com/mirabilos/tex-unicodepoints in case someone wishes to have a look (READMEs and, for unicodepoints, a PDF documentation, will follow ASAP)
@JosephWright possibly I just run make and it makes whatever it makes, although the release notes mentioned a re-arrangement of library loading to accommodate the ffi thing
@DavidCarlisle Hopefully Mojca will build for Mac as well soon
@DavidCarlisle Interesting thing with the version strings: Akira sticks to 'current' TL (so says TL'16) but Mojca/Richard Koch go for the upcoming one, so TL'17 :)
@DavidCarlisle I guess I should check l3names against the LuaTeX list (again)
@DavidCarlisle I think I've got all of the primitives: the list in the .tlg (so from extraprimitives) seems to be missing a few I know exist, plus has stray 'eTeXVersion' and 'eTeXminorversion' (not in e-TeX itself!)
