LaTeX 2020-10-01 off to CTAN

@JosephWright and only a day late :)

@Skillmonlikestopanswers.xyz The last-minute stuff held me up, but it's only a nominal date anyway :)

@DavidCarlisle Because that's what it used to be - see my comment on the question

CTAN will love me: ctan.org/incoming

I see @HenriMenke is also sending a little package ;)

@JosephWright an tomorrow latex-dev

@UlrikeFischer Yes

@UlrikeFischer I'll wait until they give me the all clear

@UlrikeFischer What did you decide about the hyprerref issue?

@JosephWright well I did take the code out that broke yesterday. I also took out a few older tests. subfigure is still in, as we should discuss it. I could make a release (or David), but if it is not pressing anymore after the latex fixes I could first try to get this greek numeral stuff out (it should go into babel-greek).

@JosephWright I didn't benchmark yet (and wrote it with a slightly different syntax), but I think I have a faster \__tl_act:NNNn code.

@UlrikeFischer (@DavidCarlisle) On another topic, I was thinking about zref, and wonder if (in HO's terms) properties really should be 'shared' between lists. One could go for

\ref_list_gset::nn { <name> }
  {
    <prop1> = <code1> ,
    <prop2> = <code2>
    ...
  }

@UlrikeFischer Cool

@Skillmonlikestopanswers.xyz Good stuff: I guess we are still waiting on final decisions on the speedups you've suggested so far (I'm leaving to Bruno, Frank, ...)

@UlrikeFischer (@DavidCarlisle) I think the only property that gets reused a lot is page, and really a few page = \thepage lines might be safer overall as then each list is independent

@JosephWright er not sure I'm following the zref or your update enough to answer without more context

@DavidCarlisle :)

@JosephWright on the catcode 15 thing yes I am just slightly surprised it didn't come up more often before 2018 that \ifxetex ^^^^...\fi gave errors in pdftex, but perhaps it did and I wasn't paying attention

@JosephWright there is not only page. abspage is reused too (I reuse it ;-)), savepos I would guess too, counter. And some of the properties you find in the various zref modules have quite complicated definitions. If they can not be shared people would have to duplicate the code if they want to reuse it.

@DavidCarlisle In zref, there are a series of properties: things that are trackable, like page, abspage, xpos, etc., and they go in lists, so the 'main' list contains page and default (\@currentlabel)

@DavidCarlisle :)

@DavidCarlisle I guess it was not very obvious, or widely used

@JosephWright yes followed that far.. but not sure I see what you are suggesting changing with the list gset above

@UlrikeFischer I guess I was thinking things like savepos would be in a dedicated list you'd always use, not in multiple lists

@DavidCarlisle I guess not so many people actually used the "use raw encoding" command. (I wrote Ross yesterday night with a simpler implementation to reset the encoding, I guess that's why he commented).

@DavidCarlisle Well the current zref approach means that there is one global namespace for properties, so if package X sets up say page differently from normal for it's own reasons, it affects everything

@UlrikeFischer sure but prior to 2018 they didn't need to use that to get the error

@DavidCarlisle Indeed, but I guess there was less use of the Unicode engines

@JosephWright oh OK i see

@DavidCarlisle It's probably OK, but I want to at least think about it; @UlrikeFischer probably has a better idea about reuse of properties

@JosephWright if I would write some command or environment where I'm interested in savepos and perhaps a counter of the enviroment, I would internally define a small list containing both, and use an internal label by this list. It keeps the label short if you stick to the attributes you need, and I like it that there is "store" of useful attributes predefined in the various zref modules to pick up from.

@UlrikeFischer OK, I'll stick with the current approach and probably do an initial checkin today

oh pict2e update. Where the changes merged?

@JosephWright yes, the problem is that after the first review of Bruno I had to make some changes, and did additional stuff, so that another review (at least in parts) might be due.

@DavidCarlisle did you find time too look on my question

?

folks -- I am so frustrated that I woke up at 3.00 in the morning. I am just wondering why is it so bloody difficult to format the table of contents in the article class? I am not asking for fancy stuff like column/row shading in RGB colours, all I ask for is simple borders for the table (normal \hline etc) and I can't find a single example anywhere in the interwebzzz... \end{rant}

@VinayPandey I woke up by that time and had a cookie.

haha... were you dreaming about tables too?

@VinayPandey No, just about cookies. :)

latex makes complex things simple, but simple things impossible :-(

well, I was dreaming about table of contents and its borders

@VinayPandey the article class is very old. And at the time memory where scarce and there wasn't space for fancy configuration options. For a frame I would probably put the tableofcontents in a breakable tcolorbox.

so which class should I use which can allow me to put borders in table of contents?

I am a complete beginner

Also, fyi, I have posted my question here: tex.stackexchange.com/questions/565023/…

also @UlrikeFischer I am not looking for any fancy colouring, all I want is my table of contents to have black line borders

@VinayPandey i don't know one, imho frames are rather unusual, or only for small minitoc. The etoc package has perhaps somethings (but is imho a challenging package for a beginner ...)

when you say <<frames>>, you mean? I am really sorry, my only experience with LaTeX has been to use pre-defined classes provided by my faculty. I will really appreciate if you can go through my question once, please!!

@UlrikeFischer perhaps hook a TikZ frame (remember thingy) to the table of contents?

alright guys, I am going to drug myself to sleep and try to get another 4 hours of sleep (it 3.45AM here); I am very very thankful to you for your help

@VinayPandey oh you mean you want to pull all the toc content in small cells. Well basically you then need to format it as longtable. But better reconsider the layout, that doesn't look good.

@UlrikeFischer so for example, I can do \begin{tabular}{c|c|c} blah & blah & blah \hline \end{tabular} and it will give me all the table borders

is it possible to do something like that for table of contents?

for example \tableofcontents[borders=true] or something like that

and also, how do I reformat a tableofcontent as long table?

really?!! no one has ever asked for this?

opps, deleted by mistake

Here is it again

@VinayPandey Your image is nor particularly clear as to what you want to do. As Ulrike says, this is a very unusal request, I;ve used latex for more than 20years and never seen it before, so no there is no easy solution for this other than completely reimplementing how the toc is formatted. You are also forgetting what should happen if entries are longer than one line? You table example just now does not support that. So this is not an easy question

@VinayPandey I've never seen it before. It looks like something one had done in Word in the 90s (but that is just my opinion)

I am creating a court document, I don't think they care about aesthetics

@VinayPandey not many. But the etoc manual shows a toc of this kind.

this is depressing

@JosephWright Just realized that my faster variant will only work for braces as begin and end group character.

@Skillmonlikestopanswers.xyz We have to normalise charcode anyway: is that an issue?

@JosephWright The current implementation works for arbitrary character codes.

@daleif while you are at it, do you mind answering the second part of my formatting question?

@JosephWright basically my speed enhancement exploits #1# type arguments.

@Skillmonlikestopanswers.xyz Ah

@Skillmonlikestopanswers.xyz I think we have to keep that

@JosephWright will still try to get something done

@Skillmonlikestopanswers.xyz Cool

@VinayPandey Note you should not use \center as a macro, use \centering for that

And as it has been say many other times, never add formatting into a macro that is suppose to just contain a string. That is bad programming.

Point 1 can be done, but it will need to serious reprogramming, especially also to get support for hyperref, so I'm not sure if this is even worth it. Regarding the header for \tableofcontents as mentioned formatting has no place inside \contentsname, you are better of copying the definition of \tableofcontents from article.cls and redefine that. Just use your code instead of the section command that \tableofcontents has.

@JosephWright what would you guess is the mean ratio of token types N, group and space in the argument to \__tl_act:NNNnn?

@Skillmonlikestopanswers.xyz Massively toward N-type; may 6-to-1 to spaces

@JosephWright and groups?

@Skillmonlikestopanswers.xyz Very rare in most texts

@JosephWright 60:10:1 or even rarer?

@Skillmonlikestopanswers.xyz That sounds about right

@JosephWright I'm just asking, because testing with is_group and then is_space is faster than with is_N_type and then is_group in the group and space cases. Time comparisons of the changed test to the order taken by the current tl_act are 1.134 for N-type, 0.899 for space, 0.41 for group. With a ratio of 60:10:1 the current order is faster in average.

@Skillmonlikestopanswers.xyz I guessed something like that

@JosephWright But luckily, I'm a smart guy, and identified two tokens which can't be inside the argument that I can use to set up a much faster test :)

Have I told you that updating documentation is so time-consuming? :-(

@yo' no, you didn't :)

@Skillmonlikestopanswers.xyz ok, then: Updating documentation is so time-consuming!

@yo' What was the worst advice you had to fix?

@samcarter_is_at_topanswers.xyz \hfill\break might be a good candidate. Usage of \mathlarger for integrals might be another one.

@yo' outch :) Good thing you are doing the clean up!

@samcarter_is_at_topanswers.xyz yeah. But it's so much work, because you have to update the code, create the screenshots, correct the example document, make sure that everything works, and all that not in LaTeX.

@yo' Ohh, poor yo' !

@samcarter_is_at_topanswers.xyz :)

@yo' while you are at it: could you replace the Compile button by a Typeset one?

@PauloCereda not gonna happen I'm afraid.

@yo' yes, creating the screenshots was the most tiring, imho.

@Skillmonlikestopanswers.xyz and getting the "Open in Overleaf" links to work.

@yo' those were fine for me, most of the times at least :)

@Skillmonlikestopanswers.xyz today I saw a bunny

@PauloCereda was it tasty?

@Skillmonlikestopanswers.xyz oh no

@Skillmonlikestopanswers.xyz they tend to stop working (API inconsistencies)

@yo' ok, that wasn't the case one and a half years ago, at least afaik.

@Skillmonlikestopanswers.xyz I just today found one whose underlying project doesn't exist. And it's impossible to create new ones.

@yo' A ghost project that only works at midnight?

@samcarter_is_at_topanswers.xyz who knows :-/

@JosephWright is there a expl3-variant of \NoCaseChange for the \text_uppercase:n somewhere? I know I can create one with \l_text_expand_exclude_tl etc, but it seems a bit like a waste if everyone does it (I need it to avoid that text_uppercase uppercases a \msg_warning argument ...).

@UlrikeFischer I didn't create a 'standard' one: probably one for the team list, but adding a standard name is trivial

twitter.com/AbbieEvansXO/status/1311753301721968640

> yes, sharks can outswim you. but you can outrun sharks. so far in a triathlon you’re square. all comes down to who’s the faster cyclist

bike sharks wooo

@JosephWright 10% faster \tl_reverse:n for 7575 tokens of input (ratio 60:10:1, each group containing 3 tokens). And 19% faster for 75 tokens of input.

@daleif @daleif where do you suggest I should put the formatting?

@UlrikeFischer@DavidCarlisle any ideas?

@JosephWright I can't run the test suite :(

! Critical LaTeX3 Error: Mismatched expl3 detected! Loading expl3 will abort.
(LaTeX3)
(LaTeX3)                 The version of expl3 preloaded in LaTeX is
(LaTeX3)                 ....-..-.., but the version of expl3 you are trying
(LaTeX3)                 to load is dated ....-..-... This is most likely
(LaTeX3)                 due to a stray format file in the user tree. Please
(LaTeX3)                 make sure that your LaTeX format is updated.

@Skillmonlikestopanswers.xyz You likely need to merge master into your branch (your branch contains an older expl3 than the one loaded into your LaTeX format)

@Skillmonlikestopanswers.xyz Or try undefining \__file_mismatched_dependency_error:nn, but I didn't tell you that ;-)

okay,

\footnotesize is looking good for my need.

@VinayPandey as I mention in my comment on the question, see how tableofcontents is defined in article.cls, make a copy to your class and replace the section part of it with the definition you were adding to contentsname

@daleif which is exactly what I said yesterday...

@Maïeul no busy at work will try to look over the weekend (if I can find it again:-)

@VinayPandey that is why I changed \center to \centering in my answer to your first question

ah, I was trying to do \centering{stuff} and that was centring everything after stuff as well. I changed it to \begin{centering} stuff \end{centring} and that fixed it

this is great. thanks!

@VinayPandey No!!! \centering is the command form, center is the environment. \centering doesn't take an argument so {\centering stuff \par}

so... how did begin/end centering fix it?? rips out hairs

@VinayPandey that is the same as {\centering stuff} but will not center unless there is a paragraph break. There are several answers on site about this. Best to slow down, stop ripping hairs and ranting...

my apologies

@UlrikeFischer I might need a little while to look at the Greek stuff

@JosephWright I found something that worked (by using two commands, one for text one for the bookmarks. But it is not symmetric, it only works for uppercase, not lowercase, as I can't setup the replacement command differently for both cases. I will sent it too you. (And don't hurry, that is not pressing, only interesting).

@UlrikeFischer OK: I take it in the demo I need to uncomment the 'bad' lines to see the issues

@JosephWright yes, in the first file you should uncomment the \textuppercase lines. In the second \textlowercase simply doesn't lowercase the numerals.

@PhelypeOleinik great, now I can finally not run the test suite... :)

! Undefined control sequence.
\__debug_generate_parameter_list:NNN ...tl_set:Nx
                                                  \l__debug_internal_tl {\ex...

@PhelypeOleinik (most likely you're not the right person to address): The footnote on the first page of the firstaid documentation is empty.

Totally off topic, if you need some entertainment, the absolute king of mashups just dropped a new album: soundcloud.com/neilcic/mouthdreams? ref=clipboard&p=a&c=0, if you're not familiar with Neils work look up T. I.M.E., one of my favourites of the last album a mix of the vocals of YMCA and the music of Hans Zimmer, that lyrics hits so much harder this way.

@PhelypeOleinik fixed the Undefined control sequence error.

@daleif ooh

@DavidCarlisle btw, I have finally learnt how the Overleaf API that learnlatex uses works :D

@daleif I like mashups, thanks for sharing. I like Mashup Germany. :) Have you ever heard this? dailymotion.com/video/x60p7t7 Journey's Don't stop believing meets Miley Cyrus' Wrecking ball. :)

@Skillmonlikestopanswers.xyz Only to get other errors :D (please note that doing the undefining thingy I suggested may only delay another error)

@PhelypeOleinik did the load order for l3debug change somehow? The undefined control sequence error shouldn't ever happen there, as it should be called after l3tl was loaded and that one defines \__kernel_tl_set:Nx. My first quick fix for this did indeed fix that error, but then scope checking didn't work correctly.

@PhelypeOleinik No, I merged master, and then things went downhill.. :)

@Skillmonlikestopanswers.xyz Not sure how we handle post-release fixes like this. I'll let Frank know, thanks!

\begin{chat}

@Skillmonlikestopanswers.xyz No changes that I recall... l3tl is one of the first things in the kernel, so if l3debug comes before that a lot of stuff won't work.

@yo' the one at latexcgi.xyz is similar (more restricted in some places, more features in others:-)

@Skillmonlikestopanswers.xyz Cool :-)

@DavidCarlisle yeah, that's always. Which "extra features" you have? Lemme guess: easy access to all aux files :D

@Skillmonlikestopanswers.xyz Definitely not my fault

@PhelypeOleinik only post release fix mechanism we have is to release a PL1 format (what did you break?)

@yo' actually not that (not in the POST API)

@DavidCarlisle Not me, of course. @Skillmonlikestopanswers.xyz broke a footnote in the documentation of firstaid

@yo' I have parameters to specify bibtex/biber and makeindex but you probably don't need that as latexmk figures it out, also i can specify platex and zzz-dev formats in teh API (platex works on overleaf but you have to do it via the latexmkrc file, which is OK

@PhelypeOleinik if it's doc only fix in source and leave for the next release most likely

@DavidCarlisle well, platex is not OK, and I still don't understand why we don't have a native support for it.

and re:biber, yep, when latexmk behaves, it's fine. But sometimes it misses some needed steps, and then it sucks.

@yo' you can put %!TeX platex comment at learnlatex and it submits the right thing to overleaf so it works OK, just means the latexmk file is explicitly in the project

@yo' also arara. :)

@DavidCarlisle yeah, this is the standard workaround.

@PauloCereda quack!

@PhelypeOleinik ^^^

@PauloCereda make it independent of Java, and it might be doable.

@matt ooh quack

@PauloCereda what does that even mean

@PhelypeOleinik Yeah, but somehow \__kernel_tl_set:Nx isn't defined while l3debug is executed...

@yo' currently if you use %!TeX pdflatex-dev at learnlatex then that is used at latexcgi but not yet at overleaf I could do as for platex and conjoure up a latexmkrc fiel to us ethe dev formats but I haven't done it yet

But then, you'd have a latexmkrc file to fool latexmk into running arara, and then arara directives in the main file. Is it worth in?

@DavidCarlisle well, I'm not even sure which versions of -dev binaries you would get...

@yo' I think @PauloCereda meant drop latexmf and use arara instead

@Skillmonlikestopanswers.xyz :-)

@DavidCarlisle I just meant "pick the tool you want". :)

@DavidCarlisle ah that would be so much trouble!

@Skillmonlikestopanswers.xyz Don't you have an outdated (outdated == doesn't have your l3tl changes) format?

@yo' that'sthe thing you get the ones from whatever early tl2019 snapshot you used so it's a bit arbitrary and while it technically works probably isn't what anyone asking fro a dev format really wants so I didn't do it.

@yo' I don't think @PauloCereda cares if it causes you trouble.

@DavidCarlisle oh what a naughty selfish duck

@matt well, quack is the word used to describe the process of creating a word that phonetically imitates, resembles, or suggests the sound that it describes, in this case, a sound a duck makes. :)

@PauloCereda perl works well

@PauloCereda i am aware of that

@PauloCereda we should really make a duck conlang or something

@yo' oh

@DavidCarlisle there's at least some discussion to make the TL image available to beta users as soon as it's first deployed, which would be great.

@matt no make duck soup

@DavidCarlisle oh no

@DavidCarlisle never

@PauloCereda im with you :)

@yo' oh no

@matt ooh support

@DavidCarlisle ^^

@PauloCereda happy quacks

@PauloCereda yeah. We're glad we (at least on Support) know how to configure latexmk to do the right stuff. That we know how to run qpdf from latexmk. That we know all the dirty trick and all the cases where it fails. I don't think we wanna go through this process again, and rely on Java at the same time :)

@matt ooh happy quacks indeed

@yo' but you will be so far behind latexcgi.xyz (which is up to date as of now) actually tlmgr is on update 82 of 107 :-)

@PhelypeOleinik shouldn't be an issue while running l3build check in l3kernel.

@PauloCereda infinite happy quacks lets just wait for the humans to toss us some bread

@yo' I personally don't think Java poses any harm more than Perl already does. :)

@DavidCarlisle IoT images? :)

I think I'll call it a day and take another look tomorrow.

@matt ooh they could give us cookies too

@PauloCereda I ain't got room for docker images.

@PauloCereda that's even better

@DavidCarlisle UK-TUG money? :)

WE ARE THE FEDERATION OF DUCKS. SURRENDER YOUR COOKIES IMMEDIATELY

reminds me of the borg tbh

@PauloCereda if I don't need to then I'm not sure if I want to spin up a new docker image every time someone tries a two-line my first document example

@Skillmonlikestopanswers.xyz Hm... That's correct, yes. Not sure what's happening then

@yo' I think you would enjoy joining me on my morning jogging: I encounter about 40 Guineafowls!

@PauloCereda /dev/root 7.7G 7.1G 669M 92% /

@Skillmonlikestopanswers.xyz That's a good idea. Enjoy your non-debugging activities :-)

@DavidCarlisle cache :)

@DavidCarlisle this machine seems to be running on a USB stick :)

@PauloCereda because of course usb sticks can run latex

@PauloCereda it's a free amazon usb stick, but it's doing between 100 and 400 documents a day at present

@DavidCarlisle ooh colonies

@PauloCereda so many ooh's

:-) Wait for it...

@matt oh

@PauloCereda there we go! :)

@matt 8670

There it is!

@DavidCarlisle OH MY GOSH THAT IS A LOT OF OOHS

ooh a counter

@PauloCereda I raise you Bustin: youtu.be/0tdyU_gW6WE

@PauloCereda 8671...

@daleif LOL

@daleif cool!

That one could even have charted

@daleif ooh a youtube link

Btw here's T. I. M. E. youtu.be/DsoCe7C4Kmk

ah yes, time

@matt I think Paulo have exclusive rights on "ooh"

@daleif w- but- eh fine

Lol hadn't seen the previous discussion

On ooh

@daleif ooh

> @matt I think Paulo have exclusive rights on "ooh"

@PauloCereda is this true?

@barbarabeeton, @egreg, @yo' ^^

@PauloCereda why

@matt ooh a conundrum

@PauloCereda i'll take that as a yes

now i have to do some factoring ugh

if only latex could help with that

@PauloCereda I should probably talk to you about that at some point (I have a similar setup where I'm going to need docker or something) but if you have 10 users all running examples you need to instantiate 10 docker images don't you, you can't avoid that by caching the first unless I miss something?

@barbarabeeton no, not at all. But it struck me when I had 2 days to work on documentation and did much less than I wanted... :-(

@yo' -- You know the formula for estimating how much time a task will take: (1) Make your best guess. (2) Double it. (3) Increase the order of magnitude by one (i.e., hours ==> days). Happy estimating.

@barbarabeeton if i think it'll take 30 minutes, it'll take 60 hours?!

@matt -- Yes. (Or you could say 1/2 hour ==> 1 hour ==> 1 day. Hmmm. That;s not the same thing. Better be careful when you present an estimate like this to your boss.)

@barbarabeeton math is weird sometimes

@matt -- Yup.

personally 1 day sounds nicer than 2 1/2 days

dont know about you though

@barbarabeeton -- This is great news for the Earth and global warming... We have much more time than those silly scientists let on.

I like @DavidCarlisle's duck jokes.