2:57 AM
@ChiefO'Brien -- for some definition of "polite".

4 hours later…
7:20 AM
@ChiefO'Brien I guess you're not British, right? :)

2 hours later…
8:57 AM
quack

:55224180

@PauloCereda good!

9:32 AM
@PauloCereda the last split second is missing, which is pretty unsatisfying!

@UlrikeFischer wave tided up; I might talk to Uwe Kern about cmy as there is somehting I don't get

@yo' is that option now available to regular users? It used to be only available to support staff, iirc.

9:54 AM

@MarcelKrüger ooh a blog

@DavidCarlisle can html handle the french quotes spaces? loin »?

10:31 AM
@UlrikeFischer define handle

@DavidCarlisle avoid a line break.

@UlrikeFischer if you want a thin space rather than a normal word space you'd need &thinsp; or similar.

@DavidCarlisle the door thingy you use to open and close and stuff
oh wait

@PauloCereda In an emacs buffer on this screen it's a pointer to a pointer that hopefully has something allocated by something earlier as it's about to get accessed....

@DavidCarlisle ooh a pointer to a pointer

10:35 AM
@DavidCarlisle I didn't meant the size of the space, but how to avoid that » ends on the next line. How do french writers handle this? They probably don't insert &nbsp; all the time?

@UlrikeFischer if you have a fancy keyboad isn't inserting U+00a0 as easy as inserting U+0020 just as é is as easy as e ?

@DavidCarlisle I don't know if a french keyboard has a key for this, but the markdown code of the french translation only uses U+0020, that's why I was wondering if it will work ok.

@UlrikeFischer No, we could (probably should) fix the markdown or it can be fixed in javascript but that causes a reflow for the readers view of the document:
2

In French typography conventions, few punctuation characters like ;,: or ? must be preceeded by a white space. This is causing unwanted line breaks when the punction is at the border of its HTML container. Voulez-vous coucher avec moi ce soir ? Non, merci. becomes Voulez-vous coucher avec moi...

@UlrikeFischer actually we could fix it in jekyll but that would possibly noticably slow down building the site (run a regex replace over all the text replacing space[punctuation] by nbsp[punctuation]
@UlrikeFischer which of course someone has already done: sylvaindurand.org/improving-typography-on-jekyll

@DavidCarlisle ;-)

10:52 AM
@UlrikeFischer, @DavidCarlisle You remind me we need to work out what to do about colour in LuaTeX: do I implement as luacolor-like for that engine (basically a backend adjustment, so transparent other than helping with spacing)
@UlrikeFischer Probably won't be too hard to set up luacolor like work in l3color, as there's no need for the complex detection stuff, although it does remind me at some stage I have to think about what to do about l3pdfmode.def for LuaTeX. (Do we perhaps want l3pdfmode-luatex.def and l3pdfmode-pdftex.def, both loaded by l3pdfmode.def?) (@DavidCarlisle)

11:19 AM
@JosephWright if we can make it more or less transparent to the user I think that would be good, handling colour via specials is such a pain. So even if it causes some engine incompatibilities it may be worth doing

11:47 AM
@DavidCarlisle It's easy enough to re-use Heiko's code, though it will need the hook code (so I'll wait at present)
@DavidCarlisle We are talking about new uses, so it's likely OK: I'll raise on the team list

@JosephWright I think it need a bit thinking. Using attributes is certainly better, but if I want to load l3color in hyperref to use the export functions, it is perhaps not a so good idea if it changes the color handling under the hood (nn the other side I'm only doing in the experimental driver, so perhaps it it is fine - it changes stuff anyway). Also currently luacolor affects all color as it changes \set@color, I have no idea what happens if you mix "traditional" code with lua.

Speaking of Heiko's code @UlrikeFischer OK for me to push oberdiek,hopatch,pagesel to ctan?

@UlrikeFischer I'll address in a team mail

12:15 PM
@DavidCarlisle vim with italics ^^

@DavidCarlisle imho they should be ok.

12:36 PM
@MarcelKrüger is there anything that luametatex doesn't break? ;-)

@UlrikeFischer :)

@UlrikeFischer the developer's neck? :)
Ergonomics FTW

@UlrikeFischer @MarcelKrüger is brave trying to get anything working on LuaMetaTeX

@JosephWright s/get/break/g , s/anything/everything/g :)

@PauloCereda how can you write a document long enough to need longtable using vim?

12:47 PM
@DavidCarlisle oh longtable will be incorporated in the kernel, right? :)

@JosephWright Anything except ConTeXt ;)

@TeXnician you follow the ConTeXt list? I'd say anything is accurate :)

@JosephWright ooh

@JosephWright Actually, yes. But many of the reported breakages would have not been unlikely for MkIV if that would have been continued (modulo engine changes which ConTeXt's higher levels try to hide)…

@TeXnician Sure, true

12:57 PM
@JosephWright Probably it's more about ConTeXt being a moving target and LuaMetaTeX is now more considered as part of ConTeXt than LuaTeX has ever been.

@TeXnician Well yes, and of course Hans needs ConTeXt to work for his day job

@JosephWright And that's probably what makes ConTeXt quite usable apart from all the rough edges. Some things are just working where LaTeX is still struggling. But I understand that Hans' strategy is obviously something that makes generic programming very hard.

@TeXnician Hans' position is very clear on LuaMetaTeX, so I'm not complaining, and it's well-known that I'm often the 'ConTeXt fan' on the LaTeX team (@DavidCarlisle)

@JosephWright Yes, LuaMetaTeX is not primarily intended as the general-purpose TeX engine but as the next step in what ConTeXt needs. Nevertheless I really appreciate that the LaTeX team still considers it a target because I sometimes love to use expl3 in ConTeXt :D

@TeXnician I hope it proves useful!
@TeXnician I don't always see the 'what ConTeXt needs' aspect to the changes, though of course Hans has big ideas; new primitives/functions, yes, changing old ones less so (removing primitives, fine if the idea is to use Lua)

1:05 PM
@TeXnician but the question is if "what ConTeXt needs" includes the needs of the ConTeXt users. I mean we were looking at the expl3 question not because of some generic idea to support context, but because a ConTeXt user reported a failure.

@UlrikeFischer I wonder it they want to be able to change case properly ;)
@UlrikeFischer See Frank's comment on GitHub (which I expected)

1:20 PM
@JosephWright yes, I saw it already and basically I agree. I mean I don't mind to play around with it to get it working but as long as it is so unstable, it should clearly be marked as "fun and experimental code", and not as something that we try to support seriously.

@UlrikeFischer There's a reason I gave up trying to build a LaTeX format with LuaMetaTeX (when it wouldn't even give an error message in IniTeX mode)
@UlrikeFischer See the demo in my email

:55226370 (removed)

1:38 PM
@JosephWright but the interesting output is there also without luacolor (and xcolor instead).

@UlrikeFischer Ah, hadn't thought of that
@UlrikeFischer Not for me: only loose red at the end with luacolor

@JosephWright somewhere there is \__color_backend_rgb:n instead of for me, \__color_backend_select_rgb:n but I can't find it ...

@UlrikeFischer I just tested on my non-development system, so that doesn't come up ...

@JosephWright I get the same issue with xcolor instead of luacolor too.

@MarcelKrüger I'll check again then: I must have messed something up

1:53 PM
@JosephWright I just tested again, it was a dev/non-dev issue: I get the right behavior with xcolor and the issue with luacolor when using the dev formats, while getting a consistent error with additional weird text when running the non dev formats.

@MarcelKrüger I've just made sure my local tree was clean and rebuilt my pdfTeX formats, and there both release and dev seem happy; are you sure you've not got a stray file somewhere?
@MarcelKrüger To be safe, testing on latexcgi.xyz: definitely works with pdfTeX loading xcolor, also OK with LuaTeX unless luacolor is loaded (@DavidCarlisle!)

@JosephWright It seems that it picked up some local l3 files, the issue is gone after rebuilding the formats.

@MarcelKrüger Phew

2:24 PM
@JosephWright I had again a local format in .texlive2020 ...
@JosephWright btw: it is rather confusing that part of the color code is also in l3color-base :-)

2:41 PM
@UlrikeFischer That's intended to be temporary: we are likely just about at the point when it can be merged across
@UlrikeFischer I think the interfaces are OK, but we should have formal review, and at the moment there is other stuff as a pririty; perhaps next month?
@UlrikeFischer Oh goodness

@JosephWright didn't really expect this to be a generic sandboxed test enviornment:-)

@DavidCarlisle No, but it's a handy method: let's be honest, the users as a sandbox = the active members of the LaTeX team

@JosephWright plus all the people who complained when we split oberdiek and broke their travis setups:-) but still only a handful of people, anyway we are not over stressed today:
Wed Aug 12 14:51:54 UTC 2020
Number of distinct users:     3
Number of document requests:  19
Number of documents returned: 14
Number of unique user agents: 3

@DavidCarlisle Hopefully that will rise once Jonas is done; we have a good amount of interest in translation, so I'd expect to see take-up from users

@JosephWright yes at which point this may be an issue /dev/root 7.7G 7.1G 619M 93% /

2:55 PM
@DavidCarlisle Er, yes
@DavidCarlisle If we get real traction (and I'd like to), then we will need some support for the back-end

@JosephWright google suggests I can find another 1G which would make a difference (that is things I can sudo apt remove without breaking everything

@DavidCarlisle Ah, right

@DavidCarlisle is that all texlive?

@DavidCarlisle Let's get users first, then worry about this; really, server time/space is cheap, so TUG/DANTE/StackOverflow (?) between them can surely cover it

@UlrikeFischer plus some extra CJK fonts for good measure:-) (no sources no doc, no backups, but full texlive 2020)

2:59 PM
@DavidCarlisle Ah, the joys of making a small TL install
@DavidCarlisle This is why I'm after a new laptop :) I do have all the doc trees!

@JosephWright here I have all doc and texlive 2016-2020 but my disk is bigger than 8G :-)

@DavidCarlisle Indeed

@JosephWright actually I'm pretty impressed with how far you can get with a "free" small machine, and if I can get it back to 1G free (which it was before I added extra fonts) then as we don't allow image upload you have to have rather a lot of documents being generated at the same time to really use up 1G disk space.

@DavidCarlisle That's good then :)

@DavidCarlisle quack

3:07 PM

@DavidCarlisle oh no

@DavidCarlisle ooh

1 hour later…
4:10 PM
ooh a crab secret

@PauloCereda eat the crab: mystery solved.

@DavidCarlisle ooh

4:47 PM
Is there an elementary way to get a single char, say the char ' - ', from another font, say sffamily?

@cis {\sffamily-} or what's elementary for you?

If I have set a font, say "EB Garamond".

But now I want the sign ' - ' automatically from sffamily ("computer modern" I think).

Is this trivial?

@cis \catcode-=\active\def-{{\sffamily-}}. Trivial but a bad idea.

@TeXnician What? :()
\documentclass[12pt]{article}
\usepackage{ebgaramond}
\catcode-=\active\def-{{\sffamily-}}

\usepackage{lipsum}
\begin{document}

Binde-Strich

\parbox{1pt}{\lipsum[66]}
\end{document}

@TeXnician eeks. you didn't try this, you are producing a loop.

4:58 PM
@UlrikeFischer Oops, you're right. Should have used \char"2d or similar :D
@cis Even if you don't create a loop by using a char command instead of - (didn't think when writing), you should never do this. Making - active is one of the worst ideas one could have in real world documents.

@cis better don't try this. If it were this easy I would have given the answer days ago. Even with correct code you will gets lots of errors. - is used in too many places where such a definition would harm, e.g. in url or tikz.

If you really want to see it in action use something like

\font\funny=cmfi10 at 10pt
\catcode-=\active\def-{{\funny\char- }}
Binde-Strich
\bye

But as @UlrikeFischer said and I previously mentioned, for LaTeX it is definitely going to break everything.

@TeXnician This way it works:
\documentclass[12pt]{article}
\usepackage{ebgaramond}
\usepackage{lipsum}
\begin{document}
\section{}
Binde-Strich

\section{}
\font\sffamily=cmfi10 at 10pt
\catcode-=\active\def-{{\sffamily\char- }}
Binde-Strich
%\bye

\parbox{1pt}{\lipsum[66]}
\end{document}

@cis Really, don't do this! That warning is not meant as a joke (the approach was meant as one). You will be breaking your document with this hack.

(with many warnings... etc)
So wath would be a good way to get a straight ' - ' not a leaning ' - ' (from ebgaramond)?
Going into the font-file and change?

2 hours later…
7:06 PM
@UlrikeFischer all three built, does this look OK for a combined announce for the ctan upload
## Oberdiek bundle 2020-08-01

No change to any of the packages in this bundle,
hopatch and pagesel have been updated and moved to sepaarte distributions.
README has been extended and converted to markdown.

Note that if you are using a minimal tex distribution (eg for
continuous integration testing) and currently install oberdiek you
may need to additionally install hopatch and pagesel if you use
thise packages, as they are no longer contained in the oberdiek
bundle.

## pagesel package 2020-08-03
separate...

and thise ->these. Looks good. One should tell ctan that the catalog of oberdiek contains a link to soul which is not part of the bundle.

@UlrikeFischer thanks

@HenriMenke ^^^ as announced

7:47 PM
@UlrikeFischer got 1 of 3 away then power to house went:-) seems to be back now...

@UlrikeFischer no all three gone now. Not sure what happened but we have overland power cables to the village, and thunderstorms.... was only off for 5 minutes

8:13 PM
@JosephWright interesting - one reason to use expl3 is xparse ;-)

8:42 PM
@UlrikeFischer Yup
@UlrikeFischer It's true of course
@UlrikeFischer He'll like the October update

@JosephWright will it change something for context user? The package will still be there won't it?

@UlrikeFischer At the mo, loading xparse other than on LaTeX is a bit of a pain; once we role it in, you won't need to worry about hacking around that