@PauloCereda dinner

@DavidCarlisle oh no
@DavidCarlisle oh no
@DavidCarlisle Can you spot the bloke? twitter.com/paulocereda/status/1230781199632797703

@PauloCereda Petra posted something very strange about avoiding having a BOM in text files the other day, I assume it was a mis translation

@DavidCarlisle oopsie

@PauloCereda seems to be missing an "it"

9:12 AM
@DavidCarlisle I fail Englische

@PauloCereda I can't understand this at all, perhaps it made sense in German "One way to remove them: ":set nobomb" in vim."

@DavidCarlisle ooh counter-terrorism

@PauloCereda threatening use of vim is terrorism, I agree.

@DavidCarlisle If you get a mo, could you look at @Skillmonlikestopanswers.xyz's pull request?

9:19 AM

@JosephWright I could try this evening but I'm not over familiar with the current xparse implementation (in my head it's still the original version from 1990s but I'll look)

@DavidCarlisle Nothing to do with xparse ;)

@JosephWright er yes xparse/keyval same thing:-)

@DavidCarlisle almost. :)
@DavidCarlisle it's just "split at active commas, split at other commas, split at active/other equals sign, trim spaces, remove one outer brace pair, feed to macro". Now that you know what keyval is, hoosh hoosh! :)

9:45 AM

@DavidCarlisle update: hyperref (53481 -> 53837) , Have you updated hyperref?
@PauloCereda Hola Paulo, ¿como está los patos hoy?

@PabloGonzálezL Hola Pablo! Los patos están bien, gracias. E usted, como estás? (sorry for the potential linguistic butchery)

@PauloCereda jeje, your Spanish is pretty good :)

@PabloGonzálezL Meta-data

9:53 AM
@PabloGonzálezL I probably fake very well. :)

@JosephWright OK, aps...I left you a message yesterday I think (regarding l3build and specialtypesetting) .

@PabloGonzálezL I have a guilty knowledge of old Spanish-spoken songs. :)

@PabloGonzálezL I missed that: can you repeat?

@PabloGonzálezL no, only some metadata in texlive because of dependencies have changed.

@UlrikeFischer I see, that's why I didn't see any changes in the repository :)

10:01 AM
@PabloGonzálezL I asked the same yesterday: chat.stackexchange.com/transcript/message/53593982#53593982

The latest version of l3build is going great :)
just one thing, I noticed that the "order" of the files affects the result of l3build install, for example:


typesetfiles = {"demopkg.dtx", "example.tex"}
specialtypesetting = { }
specialtypesetting["example.tex"]= {func = type_example}

is perfect, but if you change the order:
typesetfiles = {"example.tex", "demopkg.dtx"}

it fails, running l3build install --full --dry-run shows the correct output
- source/latex/demopkg/demopkg.dtx
- source/latex/demopkg/demopkg.ins

@PabloGonzálezL Tables are ordered, yes
@PabloGonzálezL But thta one is a bit odd, I'll see what I can do

@JosephWright Thanks, one question, now that I saw that you marked 'os_yes' as obsolete, the documentation says that you write 200 lines, but the truth is that you write 300 lines (according to what @PhelypeOleinik wrote the other day and the code .lua)

@PabloGonzálezL ^^

10:13 AM
@JosephWright thank you very much! I didn't know I should indent \group_begin: on first level as well :)
@PauloCereda but the first telescopes didn't use mirrors but lenses :(

@PauloCereda °,..,°

Das Vampir

@PauloCereda "Der Vampir"

10:16 AM
@PauloCereda and the female is "Die Vampirin".

@PauloCereda no neutral form, could say "Das vampirische Wesen" instead.

@JosephWright regarding "expanding \cite": It could be also useful for highlighting. See e.g. this question tex.stackexchange.com/q/529401/2388

@PauloCereda I'm sure you'll be fluent after Lübeck.

10:17 AM
@Skillmonlikestopanswers.xyz I will poison the linguistic heritage. :)

@PauloCereda jaja...

10:39 AM
@PabloGonzálezL Er, yes, that is true

@JosephWright :)

@PabloGonzálezL Is that checked in somewhere I can try out?
@MarcelKrüger I think I've got the popen stuff working, and it fixes the weird issue with tools and deployment!

10:57 AM
@JosephWright You can use (github.com/pablgonz/demopkg-jw) and modify line 11 in build.lua
@JosephWright But you should use l3build install --full (without --dry-run) to check your TEXMFHOME.

@PabloGonzálezL I'll check that a bit later, and probably upload a fix today if I can

@JosephWright Grateful again...

1 hour later…
12:30 PM
@JosephWright I'm sure you're glad to hear that now the comparison with expkv looks like this:
=== expkv: ===
8.66e-6 seconds (27 ops)
8.68e-6 seconds (27 ops)
8.67e-6 seconds (27.2 ops)
=== Original: ===
1.67e-5 seconds (52 ops)
1.67e-5 seconds (51.7 ops)
1.65e-5 seconds (51.8 ops)
=== Patched 2: ===
1.05e-5 seconds (32.5 ops)
1.05e-5 seconds (32.8 ops)
1.05e-5 seconds (32.7 ops)

@Skillmonlikestopanswers.xyz I'm pretty sure we'll merge, just want to get the key-val expert (@DavidCarlisle) to check it

@JosephWright well, turns out not automatically shipping with \expanded is faster if it's not needed (but a bit slower if it's needed now).

@JosephWright will look but will not have chance before this evening

@UlrikeFischer Hi, please, how many people usually are at DANTE?

@yo' Hm, I don't know. I didn't count when I was there, I would say something between 30-60.

12:42 PM
@UlrikeFischer oh so that's quite a large meeting!

@yo' compared to some other meetings we heard from ;-)

@UlrikeFischer having the meeting above ground helps

@DavidCarlisle LOL
@UlrikeFischer well, 60 would be the typical size of a conference I attended as a mathematician :)
(and also more than attendees e.g. in Rio)

@yo' although to be fair uktug was above ground this year.

@yo' 60 would be large. I think in Heidelberg there were so many, but normally it is smaller.

12:48 PM
@UlrikeFischer ok
(we're discussing how many ducks to bring in :-) )

@yo' well if you need real numbers write to the organisator and ask.

@UlrikeFischer yep, sure, we'll see what we decide on
We also discuss whether there is anything we could bring in the meeting that would interest the others.

For once the standard answer is not "update in user and admin mode" but "delete a local fmt"

1:19 PM
@DavidCarlisle and @yo' -- secret conversation, eh?

1 hour later…
2:43 PM
(removed)

Jan 25 '14 at 19:53, by Paulo Cereda
(removed)
still have not learned how to remove things?
@UlrikeFischer we should compare our answers with \tracingall. Shortest log file wins.

@DavidCarlisle OK. If you add also \ColorNthChar{23}{examination} and \ColorNthChar{8}{examination}.

2:59 PM
@UlrikeFischer also @egreg wants to play too.

@DavidCarlisle his memory for these commands are better. I knew there was something like the tl_range somewhere but couldn't find it.

@DavidCarlisle ooh things

@UlrikeFischer tick rightfully placed (and I added error checking as requested)

3:20 PM
@DavidCarlisle With much simpler macros, of course.

3:37 PM

3:54 PM
@PabloGonzálezL I've looked, but I don't see the behaviour you describe :(

4:18 PM
@JosephWright I love that commit github.com/latex3/latex3/pull/678/commits/… btw. wasn't me. I pulled that macro that way and was wondering whether I should report it, but then thought, nah, this is bound to have a reason...

@Skillmonlikestopanswers.xyz I thought I'd let you decide if you just want to rebase everything or leave the evidence ;)

4:36 PM
@JosephWright I don't intend to retouch keys, that one is a bit big for me, would take me quite some time to get its scope, and then I wouldn't be sure at all whether I could increase its performance. And I don't care whether you fix that spurious line as part of my PR or outside of it :)

@JosephWright btw. I think that if it doesn't have to be expandable, I could perhaps get a bit more performance out of it, but the current implementation could be based on expkv so wasn't too much extra thought. The problem is, that once you document \keyval_parse:NNn as expandable, you can't revert back, while the opposite direction shouldn't break much (I just hope there is no code around relying on \keyval_parse:NNn being protected).

@Skillmonlikestopanswers.xyz I'm going to hold off a release of that code until after the TL freeze (end of next week)

@JosephWright good, that way my expkv documentation can still claim being faster than expl3 by a margin (which it still will be, as I'm comparing key-setting macros, not keyval-parsing) :)

5:26 PM
> Being forgetful can be a sign of high intelligence, according to researchers from the University of Toronto.
ooh

@PauloCereda so there is hope for me!

Thanks for the link @Rmano. I might undertake the project myself in the next few years, in the form of a Wiki. Simply adding links to the manual is a good first step, but TikZ really needs a complete API documentation with all the bells and whistles of API documentation for any other markup or programming language. The number of syntaxes, commands, keys, etc. is ridiculous, and all that power just goes to waste without top-notch tools for browsing the API.

5:47 PM
@MarcelKrüger in the french group there is an error report about an invalid list tail. Do you think that this is due to the engine bug just discussed on the luatex list?
\documentclass{article}
\RequirePackage[french]{babel}
\RequirePackage{soul}
%\directlua{luatexbase.remove_from_callback("kerning","frenchb.french_punctuation")}
\begin{document}
\hl{boom!}
\end{document}

6:08 PM

@DavidCarlisle ooh more intelligence

6:25 PM
@UlrikeFischer The invalid list tail certainly is. But on current experimental versions of LuaTeX, we still get a soul error: "Reconstruction failed". It looks like soul assumes that all positioning changes are kerning between two characters, so it gets confused when the french punctuation code applies specing which depends on additional context.

6:42 PM
@COTO Have a look at the extractor scripts in the digestif project to get descriptions from the manual automatically.

7:16 PM
@JosephWright Maybe I've explained more (a thousand apologies), without making modifications to my version in GitHub, the result is the expected:
$tree$(kpsewhich -var-value=TEXMFHOME)
/home/pablo/texmf
0 directories, 0 files
$cat -n build.lua | grep typesetfiles 31 typesetfiles = {"demopkg.dtx", "example.tex"}$ l3build install --full --dry-run | grep '-'
- source/latex/demopkg/demopkg.dtx
- doc/latex/demopkg/demopkg.pdf
- doc/latex/demopkg/example/example.pdf
- tex/latex/demopkg/demopkg.sty
- doc/latex/demopkg/example/example.tex
$l3build install --full$ tree $(kpsewhich -var-value=TEXMFHOME) /home/pablo/texmf @JosephWright Now run l3build uninstall and replace line 31 by typesetfiles = {"example.tex", "demopkg.dtx"}$ tree $(kpsewhich -var-value=TEXMFHOME) /home/pablo/texmf 0 directories, 0 files$ cat -n build.lua | grep typesetfiles
31 typesetfiles = {"example.tex", "demopkg.dtx"}
$l3build install --full --dry-run | grep '-' - source/latex/demopkg/demopkg.dtx - doc/latex/demopkg/example/example.pdf - doc/latex/demopkg/demopkg.pdf - doc/latex/demopkg/CTANREADME.md - tex/latex/demopkg/demopkg.sty - doc/latex/demopkg/example/example.tex @JosephWright So far it looks the same, but when you install:$ l3build install --full
$tree$(kpsewhich -var-value=TEXMFHOME)
/home/pablo/texmf
├── doc
│   └── latex
│   └── demopkg
│   ├── demopkg.pdf
│   ├── example
├── source
│   └── latex
│   └── demopkg
│   └── demopkg.dtx
└── tex
└── latex
└── demopkg
└── demopkg.sty

9 directories, 5 files
@JosephWright the example directory does NOT exist and the example file is without extension :( ...Do you understand a little more now?

7:49 PM
@PabloGonzálezL I'm trying exactly as you've written, modulo the fact I use Windows with a few GNU tools

8:23 PM
@PabloGonzálezL OK, this must be a platform-dependent thing, so presumably it's from the file operation abstractions; going to be a bit of work

8:36 PM
@PabloGonzálezL I think I see the issue

9:04 PM
@PauloCereda Ohh, a diagram
3

2 hours later…
11:01 PM
@UlrikeFischer I pushed some luaotfload changes which, when combined with a smaller luacolor patch, allow proper interaction of luacolor with the color feature.

@MarcelKrüger oh nice. I will test asap.

11:37 PM
@MarcelKrüger hm, doesn't work yet, if luaotfload.set_colorhandler then seems not to be true ...
@MarcelKrüger now it works, I seem to have missed the last commit at first.

@UlrikeFischer That was my mistake, I forgot to push the right commit.

11:56 PM
@JosephWright are there any reasons to stick with the (comparatively) slow \tl_if_empty:nTF implementation instead the faster \if:w` test?