« first day (3574 days earlier)      last day (70 days later) » 

12:32 AM
@UlrikeFischer Could you please also notify Karl of the immediate dependencies of hopatch such that he can update (or in this case create) the .tlpsrc.
@UlrikeFischer In principle tlmgr has the ability to resolve the transitive closure of packages but it can only do so if every package in the closure correctly declares its immediate dependencies.
@UlrikeFischer See for example tug.org/svn/texlive/trunk/Master/tlpkg/tlpsrc/… on how to do this and see tug.org/svn/texlive/trunk/Master/tlpkg/tlpsrc/… on how not to do it.
5 hours later…
6:02 AM
@HenriMenke imho hopatch has no dependencies. But hycolor and so hyperref uses it. Should I update this, I will probably remove it anyway, with the new kernel functions one doesn't need it anymore.
1 hour later…
7:20 AM
@UlrikeFischer It does have dependencies
$ grep -i '\(require\|use\)package' $(kpsewhich hopatch.sty)
\expandafter\ifx\csname RequirePackage\endcsname\relax
8:18 AM
@HenriMenke that is only in the legacy package, the hopatch.sty used on current formats after the next release doesn't use that (or in fact doesn't do anything much at all) (Karl has previously said that conditional inclusions should not be listed as dependencies, only packages always used in all engines)
$ grep ltxcmds *sty
hopatch-2016-05-16.sty:  \TMP@RequirePackage{ltxcmds}[2010/12/12]%
hopatch-2016-05-16.sty:  \RequirePackage{ltxcmds}[2010/12/12]%

$ ls *.sty
hopatch.sty  hopatch-2016-05-16.sty
@HenriMenke well yes, but this is not really an independent or new one. I mean if you use hyperref you need ltxcmds anyway, and as David says, it will be gone with the next latex.
8:35 AM
@PauloCereda I refer you to a previous answer
@DavidCarlisle oh
8:53 AM
and people complain that LaTeX error messages are opaque, what could be clearer: Package inputenc Error: Unicode char ñz\endcsname@safe@activesfalse39 (U+C66) not set up for use with LaTeX
in the old days error messages had a unique numeric code
I feel that would help us these days to just google the error
@Anush You mean like Windows BSOD errors? They never helped me when I got them because I had no means to search for them when I got them (back then) :D
I have a vague of a Vax having these sorts of errors
VAX is a line of computers developed by Digital Equipment Corporation (DEC) in the mid-1970s. The VAX-11/780, introduced on October 25, 1977, was the first of a range of popular and influential computers implementing the VAX instruction set architecture (ISA). A 32-bit system with a complex instruction set computer (CISC) architecture based on DEC's earlier PDP-11, VAX ("virtual address extension") was designed to extend or replace DEC's various Programmed Data Processor (PDP) ISAs. The VAX architecture's primary features were virtual addressing (for example demand paged virtual memory) and its...
but I could have that confused with another system
9:07 AM
@DavidCarlisle some spanish special code?
@UlrikeFischer that would be my guess, : most likely,
@UlrikeFischer ooh buenos dias
@PauloCereda not enough tildes and dots in this word to trigger the error ...
@UlrikeFischer oopsie
9:26 AM
@PhelypeOleinik I'm not sure if this is still relevant, but if I hit enter here

I get an should not happen error:
! LaTeX Error: File `blub.sty' not found.

Type X to quit or <RETURN> to proceed,
or enter new name. (Default extension: sty)

Enter file name:
! Undefined control sequence.
<argument> \ERROR_should_not_happen
@UlrikeFischer ooh a blub package
I have a basic latex coding question. In bpa.st/OBZA I define myframe. Say in the second displayed box I want to change boxrule or colframe, how can I do that?
Yes, they had a unique number and a big manual to search for it. Not that the manual was too useful, once found...
The best thing of that system is that to format disks you had to confirm typing the disk label *backward*. Too fun.
@Rmano all true :)
I am assuming my question is trivial
9:47 AM
@UlrikeFischer Then why isn't hyperref a dependency if the package is only to be used with hyperref?
@DavidCarlisle As far as I can see none of your oberdiek split package list any dependencies. I'm pretty sure this is incorrect.
@HenriMenke hopatch doesn't depend on hyperref, rather hyperref depends on hopatch
@HenriMenke TL have never really had a hard line on dependencies: Karl has said as much to me
posted to main
@HenriMenke The point here is that the update does not introduce any dependencies, rather it removes some in the future
@JosephWright Yes, I know and that is the reason why a minimal TL installation is a complete shitshow.
@HenriMenke It's just not the aim of TeX Live to work like that: the default is to install everything for a reason
9:50 AM
@JosephWright It has the capability to resolve dependencies, though, so why not use it?
@HenriMenke I spent quite a while working out deps for our VM, so I do understand your point: with hindsight, package authors should have been required to list these more formally in the CTAN database
@HenriMenke I'm quite happy picking off cases as and when they come up: I know I've tightened up my own packages in terms of the documentation and will likely do the same on CTAN as I update them
@HenriMenke you are putting far too much reliance on the dependencies, they were never intended to be complete, just a mild catch for some simple cases. There is no mechanism to submit dependencies when packaging at ctan and any system that relies on just mailing Karl by hand and asking him to update some more or less undocumented dependency file by hand is not really scalable to a full distribution. ..
.. If you choose to make a minimal install then it's usually simple enough to install what you need without using the dependencies at all
@HenriMenke Like I say, the think with the hopatch update is it does not introduce any new dependencies. Moreover, it will mean that from the October LaTeX kernel update, hopatch doesn't load anything at all, which means from a TL point-of-view there are no dependencies (Karl only wants 'hard' ones listing)
@DavidCarlisle Like I said, with hindsight this probably should have been more formal (but then it also needs package authors to be careful about what are direct dependencies and what are not: you need to track exactly which packages provide which commands)
@JosephWright @DavidCarlisle Don't get me started on CTAN...
@HenriMenke There's no conditional dependency mechanism, which does complicate things
9:55 AM
@HenriMenke Karl made the offer to handle dependencies more formally only some months ago on the tex live list. In between there hadn't been many updates of oberdiek/hyperref.
@JosephWright Maybe something to talk about at a future user group meeting.
@HenriMenke I'm intrigued: other than being a bit formalised, they are always OK with me (when Robin was in charge, things were more flexible but it didn't scale well)
@JosephWright yes but needs to be at CTAN, submitting to ctan and then tracking dependencies at texlive when you have no direct control on the texlive packaging isn't a workable system
@DavidCarlisle That's what I meant by more formal: their database could have dependencies listed (though then we get into the conditional business)
9:59 AM
@HenriMenke Sure, but like @DavidCarlisle said, this data really should be part of the CTAN upload, or at least trackable there
@DavidCarlisle Why mailing Karl? There's the DEPENDS.txt specification for packages going to CTAN. Or did I miss the point here?
@TeXnician That's a new thing, it was 'mail Karl' until very recently
@JosephWright Hi, please is it right you'd still be interested in having an Overleaf blog post about learnlatex eventually?
@JosephWright Ah, right.
How can I say sth. like that:
HyphenChar={\sffamily -}, % works not
10:00 AM
@yo' Oh yes definitely, once the design work is done
% arara: xelatex

BoldFont=EBGaramond-Bold, ItalicFont=EBGaramond-Italic,
HyphenChar={+}, % works
%HyphenChar={\sffamily -}, % works not


\section{Enforce hyphenchars}
@cis You cannot do a font switch at that point. This is only intended to specify the character of the font you define with setmainfont which will be used for hyphenation.
@JosephWright Regarding CTAN: No version control, no clear way to fix unmaintained packages, not cryptographically signed.
@cis I think you would have to set up the font as a lua virtual font and substitute the character -
@TeXnician Mmmhh, I understand. I found a unsolvable problem.
@DavidCarlisle Should I change the HyphenChar with Fontforge?
10:03 AM
@cis everything is solvable, or you could use the post linebreak filter and change any - glyph node that is at the end of a line by one from the other font (and adjust the justification) or ....
@cis oh, or that yes (if teh font licence allows)
@DavidCarlisle Here I would probably know how to do it. Not with the other things.
@JosephWright mail sent ;)
10:19 AM
Kindly help with this issue:
@Anush Maybe I'm missing something, but why not simply overwrite the colour key:
	boxrule=0.8mm, text width=7.5cm, text height=1.5cm, enhanced,colback=white,colframe=framecolour,coltitle=comcolor,
	sharp corners,
	attach boxed title to top left={yshift=-0.3\baselineskip-0.4pt,xshift=2mm},
	boxed title style={tile,size=minimal,left=0.5mm,right=0.5mm,
		colback=white,before upper=\strut},
(you could also have a look at the tcolorbox theorems library, this would save you from manually writing stuff like Theorem each time)
10:40 AM
@HenriMenke CTAN was never meant to be an archive in the sense of a history of packages, rather it's a place to look for them: pre-dates the modern internet, really
@HenriMenke On signing, the issue there is most TeX 'developers' aren't real programmers at all, rather users making something 'handy' available, and requiring formal structures would simply stop them uploading to CTAN and we'd be back to lots of personal upload locations
@HenriMenke On unmaintained packages, the issue is that c.t.t. is no longer really active, so giving notice of a take-over is hard - I don't have a suggestion there, other than a notice in TUGboat - how would you manage it
@HenriMenke The historical archive business is partly hindsight: today with large storage, Git, etc., one might provide a history, but CTAN was set up as basically a box-of-tapes ...
@HenriMenke Probably a topic for some mailing list or at least threaded email: do you want to take this to mail?
@HenriMenke Or perhaps a dedicated room here in the first instance?
@samcarter_is_at_texnique.fr Do mean on this line? "\begin{myframe}{\textsc{Theorem} {\normalfont (Markov's inequality)}}" ?
@samcarter_is_at_texnique.fr I want to use myframe but be able to override the parameters as I choose each time I use it
sorry if I misunderstood
@UlrikeFischer It was relevant, until yesterday night ;-) (thanks anyway :)
@Anush ... you can already overwrite them each time you use it, \begin{myframe}[colframe=blue]...
10:56 AM
@samcarter_is_at_texnique.fr oh thanks! At least I was right that it was trivial :) Thank you
@Anush What I was thinking of was to not define myframe and then do the theorems manually, but you could use the library to create a special environment for theorems that inserts the words Theorem etc automatically
@samcarter_is_at_texnique.fr that sounds like a good idea
@Anush as a starting point:
	boxrule=0.8mm, text width=7.5cm, text height=1.5cm, enhanced,colback=white,colframe=framecolour,coltitle=comcolor,
	sharp corners,
	attach boxed title to top left={yshift=-0.3\baselineskip-0.4pt,xshift=2mm},
	boxed title style={tile,size=minimal,left=0.5mm,right=0.5mm,
		colback=white,before upper=\strut},
11:13 AM
thank you!
@Anush You're welcome! Have fun with beamer :)
thank you :)
12:01 PM
@DavidCarlisle, @UlrikeFischer I'll write some notes on the HINT business at least as far as primitive coverage goes: OK?
@JosephWright You mean about which primitives he should support? Sure, but as I (tried to) say: imho his main other problem is that he doesn't know yet how to implement links and colors so that it works with the reflow stuff and "jumping" into random locations.
@UlrikeFischer We can make some notes first
1:11 PM
@UlrikeFischer depends a lot on how his viewer works, reflowing links in pdf is particularly hard as you need to overlay the separate link rectangles, but in something that is closer to html they just flow essentially as an attribute of the text box
@DavidCarlisle his problem is that he hasn't yet decided how to mark e.g. colors: if he puts a special at the begin and the end, then it could be that the viewer jumps in the middle and so don't see the begin color command. The alternative is to use node attributes, but this enlarges the files. (at least that was what he wrote in his docu ;-))
@UlrikeFischer ah yes, I remember, Well that's not our problem:-)
@DavidCarlisle, @UlrikeFischer Email just finished and sent, I think the above fits in with it, but we can refine on the team list
@DavidCarlisle I wonder how he's handling the entire graphics area: SVG-based?
@UlrikeFischer I think either he needs a DOM or node-based approach: if you just jump into a file, it's going to have to have all of the data for each node
@DavidCarlisle no not really, well perhaps we should tell him that whatever he does, it would be easier if his specials uses a known syntax.
@JosephWright if I remember correctly he supports currently jpg and png (but I'm not sure).
@UlrikeFischer I meant more the drawing stuff, like 'put a line here, a curve there'
@UlrikeFischer Using \special or using some image primitive or ...?
@UlrikeFischer I'd look at dvisvgm if he doesn't want to go the pdfTeX route of separate primitives: dvisvgm is nice and clear on which other backend specials it supports and details it's own ones
1:24 PM
@JosephWright I don't think that it is already there, but picture mode should work, and what else does one need?
@JosephWright he may not care if it's calling to the android layout routines it might be like textures used to be on the mac, which was "draw this here" and the OS pretty much decided if it could do what was asked.
@DavidCarlisle I was imagining that in the end all of the reflowable ideas seem to be HTML + CSS + SVG as the viewer format
@DavidCarlisle Still needs some kind of instruction: I can live with some arbitrary commands if we know what they are
@UlrikeFischer Won't do clipping :(
@JosephWright that would make life simpler, although if it is that, generating html+css+svg might be a simpler target to cope with than a new format with custom viewers....
Hyphenation screw-up of the day: non-etheless (from paper I was reading).
@DavidCarlisle Fair point
@DavidCarlisle Still might be easiest with a new engine (Although LuaTeX can presumably do it?)
1:33 PM
Now that Ethel is gone, the non-ethelness of the group is very noticeable.
@JosephWright wow, who wrote that? (must be someone nice: "tagpdf package (Ulrike Fischer): highly capable package ...)
@UlrikeFischer Graham Douglas: he's the Overleaf TeX expert
@UlrikeFischer I asked him on Twitter about his target audience, as I can't imagine the average Overleaf user needing some of the detail he gets into
@UlrikeFischer We should talk to him :)
@JosephWright @UlrikeFischer interesting stretch image example: \image=Loewe.bmp width 135pt plus 16pt minus 16pt%
@DavidCarlisle ???
1:41 PM
@JosephWright Yes, indeed. It's very comprehensive.
@JosephWright one of the hint example docs "image that acts like glue"
@DavidCarlisle Ah
@DavidCarlisle Link?
@AlanMunn overleaf.com/learn/latex/Articles, for example the box one, are really aimed at hard-core developers I'd say
@UlrikeFischer doesn't mention picture mode at all. What sort of latex article is that?
@JosephWright Some of the titles of the articles are quite amusing. "Pandora's \hbox: using LuaTeX to lift the lid of TeX boxes" and "The stoic resilience of PDF within a digital ecosystem".
@DavidCarlisle but it has a video of my documentation ;-)
1:47 PM
@AlanMunn :)
@JosephWright it would be good if he would add some remarks about Frank's talk and the project.
@UlrikeFischer Perhaps, but his last section is mainly about 'available now' stuff
@UlrikeFischer I've asked him :)
@DavidCarlisle I hope \image isn't the long-term name for the primitive!
@JosephWright yes, also I'm not sure how many people will read until the end ;-) That is rather much stuff, even for me who knows what he is speaking about and can skip large parts. But it is good that there is someone who is interested - it will probably help with tests and examples.
@JosephWright on general principles, or is there a known clash with some package?
@DavidCarlisle General principle: surely people use \image or \begin{image} .... \end{image}
2:01 PM
@JosephWright yes not disagreeing, just checking. not to mention image and coimage of functions eg U+22B7 although that's \imageof in unicode-math it seems :-)
2:20 PM
@DavidCarlisle My feeling is an environment is most likely, but you could be right on the maths aspect
@JosephWright just looking at perusall, seems to be a pdf-social-annotation thing so even if we made a one page html, I think the idea would be to print it to pdf and upload that
@JosephWright just looked for an example but nearest is tex.stackexchange.com/a/85093/1090 so not actually clashing:-)
@DavidCarlisle Right yes
@DavidCarlisle I just think end users will have it (though that's likely acceptable)
3:19 PM
@egreg Sorry for shamelessly pinging you like this but could you give a look at my comment on this answer of yours?
3:33 PM
@campa I don't think that it is a duplicated, newtx has changed a lot in the last four years (and it looks as if makes upshape numbers by purpose.
3:51 PM
@UlrikeFischer I might agree for the dupe, but I still think that the “poor person's replacement” egreg is showing does not come from the TeX Gyre fonts.
@campa I didn't look in detail, sorry.
@UlrikeFischer I'm relieved... I've never seen you being wrong about fonts and I was seriously questioning myself :-)
@campa I was referring to the new question on the site.
\slshape 123 abc \itshape 123 abc
@UlrikeFischer Erm, what's the english for "aneinander vorbeireden"? :-)
My impression is that morerecent versions of newtx have defined a slanted font but which basically looks exactly like the italic one, except for kerning and numbers
My "closing as duplicate" was referred to the statement "how to get slanted numbers".
4:12 PM
@campa well in my opinion this is "vorbeireden" at the needs of the op. The question doesn't ask how to get a slanted font in the header, but only how to get slanted numbers. Did you thought to actually ask if "slanted" in the question refers to the font property or if italic is meant?
@UlrikeFischer, @DavidCarlisle Based on the luacolor ideas plus tagging, you might also notice I'm working on splitting up the backend code for LuaTeX and pdfTeX
@JosephWright I was a bit busy - I got a request to adapt an older file and naturally run into a hook bug. But imho splitting is the right thing to do. The differences are building up.
@UlrikeFischer I'd seen the bug report :)
@JosephWright ;-). Getting this out of a rather large class can take some time ;-) (at first I thought, eso-pic needs an update).
2 hours later…
6:14 PM
@AlanMunn -- that was user-defined. TeX's default is "nonethe-less" so even if that's not wrong, it should be in the exception list. I'll make sure it gets there.
@AlanMunn -- Aaarrrrggh!
6:30 PM
@barbarabeeton Given the authors and the journal, it's unlikely to have had a LaTeX source.
@barbarabeeton We linguists can make sense out of non-sense. :)
@AlanMunn "The past, the present and the future walked into a bar. It was tense."
I will shut up now. :)
Feb 10 '14 at 17:34, by Paulo Cereda
@AlanMunn: I asked Siri to tell me a joke. She said, "The past, the present and the future went in a bar. It was tense."
Jan 15 at 13:41, by Phelype Oleinik
Aug 1 '19 at 16:49, by Harald Hanche-Olsen
22 mins ago, by Harald Hanche-Olsen
Jul 3 at 9:48, by David Carlisle
yesterday, by Alan Munn
@DavidCarlisle We're fast approaching the day in which all chat messages will be recycled.
@AlanMunn oh no
6:45 PM
@PauloCereda A virologist and an epidemiologist walk into a bar.
@AlanMunn OH NO
7:38 PM
@barbarabeeton "someone said cookies?" twitter.com/cctv_idiots/status/1293976147185799169
8:06 PM
@PauloCereda -- I assume that video has been "assisted". (Cute little critter.)
2 hours later…
9:41 PM
@JosephWright Sure it predates the modern internet, but that doesn't mean that it has to continue doing that.
@JosephWright When you tag releases on GitHub, you also don't need to know about GPG signing, etc. GitHub will do this automatically for you using your normal account credentials.
@HenriMenke But that needs quite a bit of back-end development, including some form of secure login
@HenriMenke It depends what is needed: today, CTAN's main function is to supply TL and MiKTeX
@JosephWright I've asked about the issue with unmaintained packages before and the answer was “just submit your fix and we'll apply it, if we think it's okay”. But that is rather vague. What if the maintainer comes back and uploads a new copy, overriding the previous fix? There has to be some communication with upstream.
@JosephWright But then why not just take all of CTAN, commit it into a Git (or SVN if that's more convenient) and continue from there.
@JosephWright Also the CTAN “browsing policy” is completely superfluous in my opinion. It would make much more sense if all packages were developed and uploaded in TDS format.
@JosephWright Why no serve CTAN directly from the TeX Live SVN then? That would deduplicate a lot of work.
@HenriMenke Two issues: MiKTeX and licensing (related)
@JosephWright CTAN already has a login and assuming that they haven't screwed it up, it should be secure.
@HenriMenke That's political: TDS is very unpopular with some people
9:48 PM
@JosephWright Make a TeX Live non-free repo.
@HenriMenke The archive structure is very long standing and has a rich history
@JosephWright That's all fine and good but we can't move forward if we insist on carrying historical baggage. When I talked to Erik Braun at TUG2019 it sounded very much like the CTAN maintainers are pretty overloaded with work. Why not just get rid of some intermediate steps and make their work easier?
@JosephWright I've also brought up the issue of signing package with Erik and I had the feeling that my suggestion was very well received, but in the end the argument was the same as yours, i.e. that it's too complicated for users.
@HenriMenke Like I said earlier, with the fall-off in c.t.t. traffic, I don't know how one handles this
@HenriMenke Given that until recently you could just send stuff as an ftp upload, you can undersand the concern
@HenriMenke I have no more insight than you into the 'coal face', although the current team do seem to have made things a lot more 'formalised' than they were in the past (e.g. checking all the files)
@HenriMenke That's TLcontrib, basically
@HenriMenke I guess the worry is that all sounds fine to 'frequent flyers', who end up picking up dev skills to some extent, but what about someone who really just want's to put 'a few cool ideas from my thesis' into the public sphere
@JosephWright Fair criticism, but then again, do we really need yet another thesis.cls?
@JosephWright Recently I've been really drinking the Nix Kool-Aid and have picked up some interesting ideas about package management. Maybe I should write it all up and send it to TUGboat.
@HenriMenke Nix?
@HenriMenke I was thinking about policy, that CTAN has always been 'send it and we will hold it'
10:00 PM
@JosephWright A declarative package manager for Linux and macOS: nixos.org/nix/manual
@HenriMenke I guess in the end if someone has the time and ideas to write a new backend, they'd take it, but that's a lot of potential work (or at least I assume it is)
@JosephWright Regarding Nix: If you use MacPorts or Homebrew right now and are a little frustrated by them, you should totally have a look at Nix. A DevOps engineer at Shopify made a video series on YouTube on how to install and use it: youtube.com/…
@JosephWright On a completely different note: Does LaTeX2e still support vTeX and Textures?
10:27 PM
@HenriMenke Er, no
@HenriMenke No e-TeX
@HenriMenke Currently no Mac (may change soon)

« first day (3574 days earlier)      last day (70 days later) »