@PhelypeOleinik OH NO

@barbarabeeton ooh twitter.com/GregoireSaunier/status/1512924636492800001

@samcarter_waiting_for_siparty oh no

@PauloCereda :)

@PauloCereda Oh yes!

@UlrikeFischer I think we need to add support for make4ht to l3build: you can't use \input ... as far as I can see, it doesn't recognise --interaction=..., etc.

@UlrikeFischer I have a working setup locally for siunitx, although most tests then fail as the outputs are all different :)

@Skillmon Having fun in Another Place, I see

@JosephWright that wouldn't matter much, I only want a number of tests in a folder so that I get a warning, when I break it completly ;-)

@UlrikeFischer OK, I'll try to tidy up properly and do a release later, so we have something useable

@JosephWright thanks. It is naturally a bit unclear how stable its log is but we will see ...

@UlrikeFischer you could also of course diff the html (@JosephWright)

@DavidCarlisle depends a bit what one wants to test. Currently I'm more interesting in error messages. How does the hyperref build call make the html?

@UlrikeFischer currently it just forces a htlatex run in specialtypesetting["hyperref-doc.tex"] = {func = type_manual} which isn't very structured really

@DavidCarlisle That's more like a PDF-based test

@JosephWright yes

@DavidCarlisle We can set that up if we want it: @MarcelKrüger's extension of l3build to flexible test approaches should make it trivial

@UlrikeFischer Indeed: I guess I'll check in 'Initial support for make4ht tests'

@UlrikeFischer, @DavidCarlisle I wonder if I should add dvisvgm and dvips routes to the standard engines too - not activated by default but available without needing to set them up

@UlrikeFischer, @DavidCarlisle For siunitx I call those etex-dvisvgm and etex-dvips, but that's probably over-kill

@UlrikeFischer So for the present we call the 'engine' make4ht? That's the easiest to use interface I thing

@JosephWright I have dvips tests (pdf based) in the pdfresources and that is useful.

@UlrikeFischer Yeah, I was wondering if we should add it to the pre-defined set

@UlrikeFischer github.com/latex3/l3build/commit/…

@UlrikeFischer That cover it for the moment?

What is the proper way to cite a question at TeX SE? Is it just to hit "share" and then to copy the url? Over at math SE they have a "cite" link below the questions, which actually pops up a bibtex entry. (Here is a link to a question just in case: math.stackexchange.com/q/5248/97236)

@mickep Just share, yes: the maths cite have more sense of wanting 'academic citablity' than we do

@JosephWright OK, so sharing a link would be fine for tugboat articles and similar then. Thanks!

@mickep Yes: that's what I'd do

@JosephWright Good.

@UlrikeFischer If you can test the l3build change, and it looks OK, I can release so we have it in TL hopefully tomorrow :)

@JosephWright I will do it in a few minutes, we are just having lunch.

@UlrikeFischer :)

@JosephWright looks fine (once I remembered to set stdengine). The dvips/ps2pdf call is already there, isn't it? At least I don't see anything special in the pdfresource build.lua, it simply calls latex and it works.

@UlrikeFischer Cool - release coming up

@UlrikeFischer Sent

@DavidCarlisle I think I'll do the xetex.def change today, see how that goes, then look at .mps later

anyone aware of an env name that would contain whitespace? (I know there's whitespace in tikz keys and beamer keys, so I'm somewhat worried)

@yo' the answer to such questions is always "yes"

/usr/local/texlive/2022/texmf-dist/tex/latex/parselines/parselines.sty:\newenvironment{parse lines}[2][]

@DavidCarlisle GOSH!!!

and this one

They are the only two \newenvironment in texlive 2022 (I did not check other definition forms)

/usr/local/texlive/2022/texmf-dist/tex/latex/unitn-bimrep/unitn-bimrep.cls:\newenvironment{TN may be empty itemize}

@DavidCarlisle ok, thanks :-)

(we're writing a new code parser for LaTeX, and being precise in things like allowed characters in env names helps clear errors like } as soon as possible)

@yo' count it as an error:-)

@DavidCarlisle that's what we're gonna do proabably :D From unusual characters, we only consider -_*+@# at the moment.

@yo' looks safe enough, only other entry in texlive is this collection

/usr/local/texlive/2022/texmf-dist/tex/latex/tudscr/tudscrtutorial.sty:\newenvironment{Trunk!}[1]{%
/usr/local/texlive/2022/texmf-dist/tex/latex/tudscr/tudscrtutorial.sty:\newenvironment{Hint?}{\tud@tut@pre}{%
/usr/local/texlive/2022/texmf-dist/tex/latex/tudscr/tudscrtutorial.sty:\newenvironment{Hint!}[1]{%

@DavidCarlisle I should define a \begin{David's lunch: 🦆} environment.

@UlrikeFischer starts that way and ends with \end{@PauloCereda} ?

Why do I not get comments like this here in chat? tex.stackexchange.com/questions/640535/…

@DavidCarlisle oh thanks a lot!

@DavidCarlisle, @UlrikeFischer xetex.def change off to CTAN: now we find out if it's safe :)

@UlrikeFischer Does mean we have multiple colour stacks in dvipdfmx, good for things like transparency

@PauloCereda -- Oh, that's a lot of fun!

@PauloCereda a new user: tex.stackexchange.com/users/267668/arara

@CarLaTeX doomed

@DavidCarlisle You are mean

@CarLaTeX oh no :)

@PauloCereda Copyright?

@CarLaTeX nope, no harm done :)

@CarLaTeX They should change their avatar to match the name :)

@barbarabeeton I did not mention eqnarray:

@StefanKottwitz I did say :)

@StefanKottwitz -- Good! Thanks for confirmation. (By the way, the book got a good review. See it in the next issue of TUGboat.)

@StefanKottwitz google.com/…

@barbarabeeton Yay!

@barbarabeeton You remind me I'm supposed to be reviewing

@JosephWright -- I really like the graphic of the critter fishing.

@JosephWright never mind, most people use google.com/search?q=https://en.wikibooks.org/wiki/… anyway:-)

@barbarabeeton Thanks Barbara! Good reviews help with tempting the publisher to accept another LaTeX book. (and some sales help too)

@barbarabeeton But no need to push sales, as the publisher is already convinced and signed a fourth book contract with me. :-)

@JosephWright -- The LaTeX Wikibook is, in many ways, just plain bad. A good idea, but bad implementation. I've given up trying to improve it; all my suggestions have been shot down by the self-selected "editors".

MOrning all

@Canageek quarter past five in the afternoon:-)

@barbarabeeton Yes, it's odd: I do wonder who's in charge

@Canageek Morning, then:-) Having fun?

@DavidCarlisle Honestly, mostly just relaxing at my partner's apartment enjoying getting to live together after being apart so long. We made cornbread ^^

@JosephWright ?

@Skillmon TopAnswers

Also found a weird thing: If you forget to tell tgterms to be Type 1, the Angstrom symbol is all messed up. But this doesn't happen with other fonts, or if you turn type 1 on. shrugs

@JosephWright which fun? Only the occasional chat, and answering answers there and here. And trying to make up for some mistakes of past-me over there.

@Canageek ah I can see being in the same counry as your partner could have advantages:-)

@Canageek with pdftex?

@Skillmon Your reply to the question from ... a while ago

@JosephWright was one such attempt. Spotted anything that's objectively imprecise or incorrect (especially the year in which expl3 became irrevocable is quite blurry to me)?

@Skillmon No, it's all fair

@DavidCarlisle because we're afraid of the longtable-man!

Gotta go! Until later folks!

@Skillmon Have A Lot Of Fun

@DavidCarlisle Yeah, seemed easiest. At work I'm writing a template for reports. Technically a package, but its just a list of alias's so I have to type less each time. I'm hoping to write a python script that automatically copies the data I need from the instrument files and generates the line I need for the report to tell the engineers what to put in their papers. (Shutter time, total experiment time, step size, etc)

@samcarter_waiting_for_siparty We should suggest it

(I also have the template graphing things for me)

@DavidCarlisle Oh no.

@Canageek so doesn't it always find type1 versions? what else do you have in pdflatex's input tree?

@DavidCarlisle Honestly, no idea from memory. I left my work files at work. But once I put in the Type 1 command, everything worked fine again.

@Canageek not sure what you mean by type 1 command (not that it matters if I understand:-)

The really annoying thing was I has the package done and working, then needed to send my laptop back to sender, and since the package was in the TeX install folders, not my documents, I didn't back it up. SO I needed to recreate the whole thing from scratch

@CarLaTeX ^^^ would look so good

@DavidCarlisle \usepackage[T1]{fontenc}

@Canageek ah unrelated to ype 1 tha's the latex font encoding not the font format type

@DavidCarlisle Ah! My mistake!

I thought the T1 stood for Type 1

@Canageek that's why I was confused:-)

@Canageek T1 == "text encoding 1" cf OT1 == "old text encoding 1" or T2 == "text encoding 2" == "Cyrillic"

Also: WEIRD there don't seem to be any TeXLive mirrors in California. I would have thought of all places that would have one, the bay area would be one. That or they aren't obvious from the URL where they are based

(It would be much more useful if they'd say where the mirror is based rather then the URL, I have no idea where most of these mirrors are, and it often makes the diffrence between a 20 minute install and a 2+ hour one)

I swear, whomever first created the "List mirrors by country, that will tell you which ones are closest" system MUST have been European where all the countries are teensy-tiny.

@Canageek vote for Californian independence so it could have its own ctan section?

@DavidCarlisle XD. I mean, my partner just left Seattle, where it isn't uncommon to see Cascadia flags, particularly in queer households

(Now that I"m dating an American, I"m learning that the US is about 50 smaller countries in a trenchcoat pretending to be one large country)

@Canageek -- the CTAN list shows one in San Jose, which is sort of in the Bay area. But the link to the mirrors page (Sites) also says it was last updated in 2017.

@Canageek The CTAN mirror list shows the city of every mirror. Especially it lists mirrors.hoobly.com in California.

@barbarabeeton Damn, I'm in Sunnyvale, right next to San Jose

@MarcelKrüger Thanks!

@Canageek -- Only 50? Well, officially. But there are areas of the state I live in (Rhode Island) that would probably be happier elsewhere. (So consider the size of Rhode Island and adjust accordingly.)

@barbarabeeton Makes sense. California wants to be at least 4 or 5

(SoCal, Jefferson, NorCal, Bay Area, off the top of my head)

@Skillmon ooh

@samcarter_waiting_for_siparty ooh

@MarcelKrüger's mirror is pretty much in the room next door (to him). :)

@UlrikeFischer oh no

@DavidCarlisle oh no

@DavidCarlisle ooh

Well, that was a mistake. I left it for an hour, and came back, and it is saying still an hour left; the last one said 50 minutes total

@Canageek It is 1 hour of "watching time". The installer feel alone if you leave. So, just sit there and look, and it will be fine.

@Canageek Probably one of the reasons why the location is not prominently mentioned: Just because a server is physically nearer does not necessarily mean that it's faster.

@MarcelKrüger It's equally mentioned along with all the other US servers; I just happen to know where UWashington is, and used that the first time as at least it is on the same coast. (I'm used to it defaulting to a university 4.5 timezones away from me when I'm in Canada, which even if it is a fast server, that is a longggg way to send packages)

(I'm just shocked that none of Stanford, UCLA, UCSD, etc etc have mirrors up, given their long involvements with open source and comp sci. UWaterloo was the best server when I was in Ontario since it was nice and close and blazingly fast)

@samcarter_waiting_for_siparty great!

I usually stick with German mirrors because it's Germany, but since my beloved FAU mirror is lagging behind some updates, I had to resort to the United Kingdom of Great Britain and Northern Ireland, namely Oxford. :)

@PauloCereda Ok, I know Europe is small, but I would have still thought something on the mainland in central or Eastern Europe would be more convenient for you.

@Canageek I used to rely on our Brazilian mirror, but @PhelypeOleinik can confirm how terrible it is. :) I think I could try one of the US ones, but since my internet speed is sort of capped, any updated server would do. :)

@PauloCereda Ahhhh, that makes sense. The UWashington and UWaterloo ones are both excellent if you ever don't want to rely on trans-Atlantic cables.

@Canageek ooh gonna definitely try them, thanks!

@PauloCereda That and the fact that 2 or 4 minutes for the daily update doesn't make that much of a difference, so I use German mirrors for the exact same reason :)

Also you think that CalTech or Standford would have a CTAN mirror, since Knuth did his PhD at one, and was a prof at the other. You know, to commemorate that.

@Canageek -- As long as Nelson Beebe is at Utah, that's usually a good bet. (If there will be a known outage, he'll announce it on the tex-live list.) Stanford, however, aside from being Knuth's home, seems to be rather resistant to TeX.

@barbarabeeton Ohh, I should try that one when I'm home in Kelowna, my American geography is weak but I think we aren't that far from Utah.

@barbarabeeton I miss Nelson, one the kindest persons ever!

Yay! I'm done downloading things! Now I can figure out what I've messed up on my laptop, as I know this compiled son my desktop before I went on vacation XD

Talking about silly things: ministryofsillyska.de

@samcarter_waiting_for_siparty ^^ :)

@UlrikeFischer tex.stackexchange.com/questions/640564/… ... suggests we might want to make the page key more complex than just 'pass to the backend'

@JosephWright well now that we have a pagecount command, we can support page=-1 ;-)

@UlrikeFischer ooh

@UlrikeFischer any guess why the code in the question failed?

@UlrikeFischer Yup

@DavidCarlisle Passes the raw value to the backend?

@DavidCarlisle I didn't check yet, I simply assumed a faulty value.

@JosephWright but it has worked previously (I thought)

@DavidCarlisle why should -1 work? Until now we didn't have any code to retrieve the page count.

@UlrikeFischer I was thinking we might want to support -n generally in the page key, which means arranging to read the number of pages for all PDF file inclusion

@PauloCereda ohh, that's silly :)

^^^ they should fix their encoding :)

@UlrikeFischer something was according to the message shown

@samcarter_waiting_for_siparty isn't that standard utf-8 output?

@samcarter_waiting_for_siparty ooh

@DavidCarlisle maybe the real challenge is how to pronounce this word :)

@samcarter_waiting_for_siparty never try to pronouce words with accents

@DavidCarlisle I wish I had you as French teacher!

@samcarter_waiting_for_siparty I could teach you now. My sister is French so I must be brilliant at it.

@DavidCarlisle we know that you are brilliant at any language :)

@DavidCarlisle yes, the normal message when you require a not existing page.

@UlrikeFischer maybe it was a feature request, I know it came up before somewhere:-)

@samcarter_waiting_for_siparty C'est vrai

@UlrikeFischer We could add the page count lookup to the bounding box search, then check if pages > 0 and if not set pages = totalpages - abs(pages) + 1, I guess (@DavidCarlisle?)

Also: typeout is a VERY useful command

Also: Has anyone else rigged LaTeX to play a noise when it is done compiling? Or am I weird that way?

@JosephWright if it adds any measurable overhead it's probably not worth it it clearly hasn't come up that often, and if you expose a page count call, somone can always do it explicitly as two calls, one to get the page count and one to get the last page included

@DavidCarlisle I can see that - I think you need two calls for the case where you have different media boxes for each page, as you need to know the absolute page number before looking that up

@DavidCarlisle I'm not saying we should make a change, just pondering things: I agree it's a pretty unusual case and can be handled as @UlrikeFischer has shown

@DavidCarlisle, @UlrikeFischer What's funny is that this is a new feature out today

(OK, it's been possible with primitives for some time)

@JosephWright I was quite keen to answer because of this ;-)

@UlrikeFischer :)

@UlrikeFischer Does show we are trying to respond to user needs :)

@UlrikeFischer I wonder if we will see something about colour stacks and (x)dvipdfmx tomorrow then

@JosephWright yes, but easily only with pdftex and luatex. Our code also works with the dvi based route.

@UlrikeFischer It's relatively easy in XeTeX too - \XeTeXpdfpagecount - but yes that's the entire point of us wrapping stuff up, same as \pdf_uncompress: for example

@UlrikeFischer Ours caches the value, too, so there is only one hit for speed

@PhelypeOleinik revtex is eating bookmarks ;-(

@UlrikeFischer can't be my fault

@UlrikeFischer Blame the original author :)

@UlrikeFischer Link?

@PhelypeOleinik too late

@UlrikeFischer Nom nom nom

@DavidCarlisle my coffee distracted me

@PhelypeOleinik this here. It is only on the dvips route tex.stackexchange.com/q/640570/2388

@PhelypeOleinik it works again if I add \DeclareHookRule{begindocument}{hyperref}{before}{revtex4-2} but I guess on arXiv that isn't a solution yet.

@UlrikeFischer I can ask Mark to include that in... The real problem is the release ritual

@PhelypeOleinik Goat's skull?

@JosephWright And one candle for every paper on arXiv using revtex

@PhelypeOleinik :)

:60876327

@PhelypeOleinik well I'm not completly sure what is going on. So someone would have too look a bit more carefully first.

@UlrikeFischer Hm... Yeah, with hyperref's rule of “must be loaded last”, placing it before the document class is weird, at least

Anyone got any .mps files around? I've been testing with one from ConTeXt and

\documentclass{article}
\def\convertMPtoPDF{}
\usepackage{graphicx}
\makeatletter
\AtBeginDocument{%
  \edef\Gin@extensions{\Gin@extensions,.mps}%
  \epstopdfDeclareGraphicsRule{.mps}{pdf}{.pdf}{%
    \ETE@epstopdf{#1}%
  }
}
\makeatother
\listfiles

\begin{document}

Hello world

\fbox{\includegraphics{cow-brown.mps}}

\end{document}

but I'd like to try some more ...

@DavidCarlisle, @UlrikeFischer, @PhelypeOleinik ^^^ Avoids using the ConTeXt converter, at the cost of needing restricted shell escape

@DavidCarlisle Er, that is the same one ;)

@JosephWright sorry misread your comment I thought you said "cost of needing shell-escape"

@DavidCarlisle :)

you could run metapost and make some eg this example tex.stackexchange.com/a/638348/1090

@DavidCarlisle Certainly seems to work ...