« first day (3771 days earlier)      last day (1169 days later) » 

12:13 AM
@PabloGonzálezL every word is true, what better documentation could you ask for?
 
@DavidCarlisle It is the clearest (literally) :D
@DavidCarlisle Aps, have you seen a PR that is in the hyperref repo?
 
@PabloGonzálezL define "seen" :-)
 
@DavidCarlisle seen = looked at ...Is one relating to documentation.
 
12:30 AM
@PabloGonzálezL I must have seen the mail originally and just looked at the description now but havent looked at the diff yet (had no tex time this week)
Happy new year every one
 
@DavidCarlisle Time is what is always missing :(
 
 
7 hours later…
7:35 AM
@PabloGonzálezL ? Anything wrong with it?
 
8:09 AM
@UlrikeFischer I have a completely blank document (well, minus frames). Evince says "some font thing failed" (nice error message, btw)
Okular is a bit better (¿or not?)
 
@Rmano interesting. I wonder what went wrong. My local version from the upload is fine, but the one on ctan is broken.
 
8:36 AM
@CarLaTeX ooh this is the best
Where is arara 6.0?!
 
8:49 AM
$ latex small2e
This is pdfTeX, Version 3.14159265-2.6-1.40.21 (TeX Live 2020) (preloaded format=latex)
 restricted \write18 enabled.
entering extended mode
(/usr/local/texlive/2021/texmf-dist/tex/latex/base/small2e.tex
 
@DavidCarlisle you are using the wrong binary.
 
anyone else got tl2021? I assume it's just teh cygwin binaries that still say 2020?
 
@DavidCarlisle I'm just installing.
 
@DavidCarlisle For me on Linux, the pretest echoes 2021, so everything fine.
 
@UlrikeFischer ^^^ It's the one the installer installed in latex is hashed (/usr/local/texlive/2021/bin/x86_64-cygwin/latex)
 
8:58 AM
@DavidCarlisle then you can sent the first bug report? I still have 2500 packages to go ...
 
@barbarabeeton oh I love them as well!
@PauloCereda Lol
 
@DavidCarlisle Probably not done yet
 
@JosephWright yes (just mailed tl list to check) not being done would not be surprising but I'd expect it to fail not install old binaries in that case,
@UlrikeFischer sent:-)
 
Finalising my TL'20 installations and installing TL'21 pre-test :)
 
@JosephWright does renaming master to main break some connection? In all repo where I have done it, git branch -r gives a warning
$ git branch -r
warning: ignoring broken ref refs/remotes/origin/HEAD
@JosephWright it looks as if one has to set ` git remote set-head origin main
` after such a rename.
 
9:33 AM
@UlrikeFischer Something like that, yes: GitHub's own instructions are slightly different but in the same spirit
 
@samcarter_waiting_for_summer ooooh
 
@Ulrike, I have just started into pdfmanagement-testphase. Now I am looking for the documentation of the pdftex primitive interface. Where should I go?
 
@AlexG l3pdfmanagement for the core one. All modules are described in the main docu (pdfmanagement-testphase). Or ask me ;-)
 
@AlexG ooh going where your heart will take you Star Trek Enterprise
 
9:41 AM
@PauloCereda <3
 
Right, how to I add TL'21 to the TeX Live Utility ... @egreg?
 
@AlexG <3
 
This is pdfTeX, Version 3.141592653-2.6-1.40.22 (TeX Live 2021/W32TeX)!
 
@JosephWright I usually use tlmgr with the pretest release
 
@egreg Er, yes, but how do I set up texbin manually? It's been a while ...
 
9:43 AM
@JosephWright Linux?
 
@PauloCereda :)
 
ln -s <TL's original path> /opt/texbin
 
@JosephWright Normally I use a dedicate terminal window where I manually add to the path.
 
Ah
$ cat /etc/profile.d/texlive.sh
#!/bin/bash
pathmunge () {
    if ! echo $PATH | /bin/egrep -q "(^|:)$1($|:)" ; then
        if [ "$2" = "after" ] ; then
            PATH=$PATH:$1
        else
            PATH=$1:$PATH
        fi
    fi
}
pathmunge /opt/texbin
unset pathmunge
 
@PauloCereda I'd be fine on Windows or Linux, but (ironically) the 'easier to use' mechanism on the Mac is not so easy to remember how to adjust
@egreg Ah, so you don't just make the switch
I'll ask on the MacTeX list
 
9:45 AM
@JosephWright one for IoT
 
@JosephWright No, I don't want to use the pretest for normal work.
 
@UlrikeFischer Thank you. First, I will try on my own.
 
@PauloCereda I think I've already seen that code. ;-)
 
@egreg YES!
 
@egreg :)
 
9:45 AM
@egreg I learned with the best
 
@egreg I tend the other way: unless explicitly set otherwise, I use the pre-test for everything
 
@PauloCereda I don't really remember where I got that from.
 
@AlexG well I rather selfish here: questions help me to identify missing or unclear things.
 
@egreg pathmunge() is a typical shell function defined in profile scripts by Red Hat, so that might be it. :)
 
@UlrikeFischer If I get stuck I will ask you, of course.
 
9:53 AM
@egreg For the present, then, I'll go with the Windows version :) (I know how to get TeXworks for Windows to use that)
 
10:35 AM
@Ulrike My first question. In texdoc l3pdfmanagement you say that entries in Page/Resources/Properties are housekept automatically in connection with BDC marks. While this holds true for dvips (thanks to the BDC pdfmark), this is not the case for pdftex/luatex were the only interface for inserting BDC marks in the page stream is \pdfliteral, as far as I know. Or do I miss some dedicated primitive that automatically populates the Properties dict?
 
@AlexG my code does it. Actually now that you say it, I need to check luatex again ...
 
@egreg I think I might have annoyed Dick :(
 
@UlrikeFischer So does mine. I add BDC-associated OCG dictionary refs manually to the Properties dict. And I remove them manually from it on the next page, where they are not used anymore (I don't want to see OCGs in the Layers tab in AcroRead if they don't exist on the current page.) For this a have defined \pbs_zap_properties: in pdfbase.sty.
 
@AlexG with pdflatex it should work fine, but I think there is something open with luatex.
 
In pdfbase, Properties are kept track in cs pbs@props@<pagenum> and finally written out on shipout.
@UlrikeFischer How? And, how do you remove un-needed entries from Properties?
 
10:54 AM
@AlexG with labels, if you run this you can see the info in the aux, and at shipout I then use the right one. But lualatex add the marks with luacode, and there they are added simply in the stream. So I think I should open an issue ...
\RequirePackage{pdfmanagement-testphase}
\DeclareDocumentMetadata{uncompress}

\documentclass{article}

\begin{document}

\ExplSyntaxOn
\pdf_bdc:nn {Span} {/MCID~0}
\ExplSyntaxOff
abc

\newpage

\ExplSyntaxOn
\pdf_bdc:nn {Span} {/MCID~1}
\ExplSyntaxOff
abc

\end{document}
@AlexG Oh no, I was wrong, I must have implemented it at some time, it works with lualatex too ;-).
 
@JosephWright You could add an engine to TeXShop that sets PATH and then launches the compiler. I'd really like very much an interface to engines in TeXShop similar to TeXworks'.
 
@UlrikeFischer Reminds me I have to track down why opacity is added more than once
@egreg Well yes, manually I'm fine in TeXworks, I'd just like everything integrated (I only have TL'20 and now TL'21 native on the Mac; on the Windows VM I've got TL'09 onwards and use TeXworks to switch versions)
@DavidCarlisle Stats over 20 a day
 
@JosephWright now that you can test properly ...
 
@UlrikeFischer Yes: I'll log an issue
 
@UlrikeFischer Your BDCs are not associated with a dictionary object and thus no Properties dict at all. But OCGs are indirect objects and referenced at various places (Catalog) and thus need to be mapped to a /name in /Properties for use with a BDC mark in the page content stream.
 
11:06 AM
This is pdfTeX, Version 3.141592653-2.6-1.40.22 (TeX Live 2021) (preloaded format=pdflatex-dev)
 \write18 enabled.
@egreg ^^ TeXworks FTW
 
@AlexG sorry? I see /Span /l3pdf1 BDC in the stream, and << /Properties <</l3pdf1 5 0 R>> in the resources, and 5 0 o is a dictionary. What do you miss here?
 
@UlrikeFischer That is a lot easier
 
@UlrikeFischer Apologies! I forgot the second run.
 
@PabloGonzálezL @Rmano @DavidCarlisle ctan confirmed that they somehow broke the pdf during installation, they updated it, now I need to wait if texlive picks it up ...
 
11:31 AM
@Ulrike I would love to use your \pdf_bdc:nn instead of \pbs_pdfbdc:nn. But I would need to be able to pass object refs as 2nd argument, not just a dictionary. OCG dictionary objects need to be created separately as they are referenced elsewhere, in the Catalog for instance.
 
@AlexG there is a \pdf_bdcobject:nn {tag} {object name} command which takes an object name defined with \pdf_object_new. (currently documentated in l3pdftools)
 
Great, I will have a look. If, for now, I want to use my own \pdfobj interface, would that be possible too?
How to pass <num> 0 R to \pdf_bdcobject:nn?
 
@AlexG Well object names must really be an object created with \pdf_object_new. It relies on l3pdf and l3backend to get the right format for the reference, I mean I don't want code which works only for pdflatex and not with xetex or dvips. But couldn't you create an intermediary object for the reference number?
 
11:52 AM
@UlrikeFischer Something that comes out in the PDF as (see below)?
2 0 obj % created with l3pdftools
1 0 R   % created with pdfbase
endobj
 
@AlexG yes, does it work?
 
Need to try.
 
@DavidCarlisle, @UlrikeFischer Do you think the graphics-def changes for XeTeX/dvipdfmx colour stacks could go into pre-testing?
 
12:07 PM
@JosephWright can one get it only in tl21?
 
@UlrikeFischer I think so: aren't updates to TL'20 done now?
 
@JosephWright Karl wrote that not (and I hope that not, we need to get latex-dev into it don't we?).
 
@UlrikeFischer I'll wait until is is :)
 
Anyone knows what lualatex does differently with a pdf compared to pdflatex? I usually always use pdflatex. Just wanted to try lualatex on this project (much faster, yay), but at each compilation (via latexmk) my Evince crashes. It never crashes with pdflatex. Any idea what gives?
 
But he wrote "As I mentioned in passing a day or two ago, this year for the first time tlnet will continue to be updated during this pretest period. That is,
regular package updates sent to CTAN will normally end up in both tlnet
and tlpretest."
@daleif lots of things are different. But without an example it is not possible to guess what could be wrong.
 
12:12 PM
@UlrikeFischer Karl told me he can disable updates to TL2020 if needed. So a package can be frozen for TL2020 and still be updated in the TL2021 branch.
 
@UlrikeFischer might be that Evince reloads too fast and that lualatexs way of writing the PDF to disk is different
Interesting. I tried disabling the automatic reload. Then at the manual reload it crashed.
 
@JosephWright but regarding the xetex.def: as the new code is guarded by the xdvipdfmx version it should be imho quite safe. I'm using it since it is there and had no problems.
 
@UlrikeFischer Well yes, but Frank was wary
 
This is too annoying to deal with, back to pdflatex
 
@JosephWright was he? I didn't see a mail about it.
 
12:17 PM
Interestingly lualatex also gave me underfull vbox basness 10000 on all pages... hmm
 
@JosephWright as alternative one could create a graphics-def-dev package (even if the name looks as if someone has a hickup ...)
 
@UlrikeFischer Yes, but I'm also mindful of adding more dev packages for one-offs
 
@JosephWright yes. Then perhaps what Paulo suggested: update only for tl21, Karl wrote that this can be done by adding a note in the upload: "If a package update should only be in the pretest, and you haven't already told me, please say so in the CTAN upload notes".
 
@UlrikeFischer Ah, right: I'll do that then
 
12:32 PM
@UlrikeFischer Unfortunately, \pdf_object_write:nn & friends don't support generic objects, only dict, array, stream, fstream :(
(unlike the \pdfobj primitive and \pbs_pdfobj:nnn)
@JosephWright ^^^^
 
@AlexG But generic streams are not available in all routes ...
@AlexG How do you handle dvips?
 
@JosephWright I need it (temporarily) for pdftex only. I know, ps2pdf don't support them either.
(see @Ulrike's and my conversation above)
 
@AlexG right, I think that came up once. The question is if one needs it. For a one shot you don't need it: you could reserve the object with \pdf_object:new, then split up the number from the reference and write it with the primitive:
\ExplSyntaxOn
\pdf_object_new:nn {myname}{dict}

\edef\myref{\pdf_object_ref:n{myname}}
\show\myref %5 0 R
%some split up command to get the 5
\ExplSyntaxOff
\immediate\pdfobj useobjnum 5 { blub }
 
12:47 PM
@UlrikeFischer +1
@JosephWright \msg_error:nn :P
 
 
1 hour later…
1:50 PM
@daleif I tried with a small test file, but I can't see it --- my evince works the same (Ubuntu 20.04).
 
The Island of TeX proudly presents the newest and greatest version of arara, the first release in the 6 series, coming to CTAN today! Check our changelog: islandoftex.gitlab.io/arara/changelog
5
@JosephWright oopsie
 
@JosephWright I just read it too ;-)
 
@Rmano this is a 300 pages book and my evince is from 18.04
 
2:46 PM
@daleif Hmm... I checked with a small example, so maybe it does not scale. Continuos visualization is a PITA anyway --- when compiling the circuitikz manual the result is that most of the time you have grey refreshing pages. In my experience okular manages better (a pity its antialiasing is simply horrible).
 
3:19 PM
@Rmano So far I'm fine with Evince. It would be nice if one could send it a refresh signal like one could with Xpdf, then latexmk can reload it when compilation is done.
 
@Rmano @daleif I use evince and it seems to refresh PDF's automatically...
@Rmano please forward this link to your daughter: twitter.com/Pokemon/status/1365319952153083910 :)
 
@PauloCereda Will do!
 
@PauloCereda This writeup is much better: islandoftex.gitlab.io/arara/news/release-v6 ;)
 
@TeXnician ooh
 
 
1 hour later…
4:39 PM
@PauloCereda it does but it seems that can be switched off using a gnome setting.
 
@daleif Really? In Fedora it is enabled by default.
 
@PauloCereda as well in Ubuntu
 
@daleif Phew!
 
I found this somewhere "gsettings set org.gnome.Evince auto-reload false" not 100% sure that it works
 
[paulo@cambridge ~] $ gsettings get org.gnome.Evince auto-reload
true
@daleif ^^ quite!
 
4:42 PM
For longer compilations it would be nicer if it waited for a signal
@Rmano I'm not too fond of the anti aliasing in Evince either. Though it might be my third monitor that is a bit old
Interesting at home my Evince does not die with lualatex. Should be the same version of Ubuntu
 
5:10 PM
@PauloCereda so with that out of the way perhaps time to ask:
Jan 12 '16 at 12:15, by David Carlisle
@PauloCereda finished your thesis?
 
@DavidCarlisle oh no
/quacks in despair
 
So, pdflatex complains that "Font shape OT1/cmr/m/scit' undefined (Font) using OT1/cmr/m/sc' instead on input line [...]", when I'm perfectly okay with that substitution. I thought I could shut it up by using "\DeclareFontShape{OT1}{cmr}{m}{scit}{<-> ???}{}" but I apparently do not understand how tex names fonts and have no idea what to put as "???". "cmrmsc10" doesn't work.
 
Any suggestions about how best to ensure obligatory components of \frontmatter are included? I'm thinking about simply adding a flag inside each one and then checking them when \mainmatter is issued. Are there more elegant ways?
 
- Hey TeX.sx people, I have a problem with package <pkgname>
- Did you read the manual?
- Er... yes
The code:
 
@AlanMunn initialise \@author to \PackageError{zzz}{GIVE YOUR NAME NOW} then if people don't use \author{... they get reminded
 
5:21 PM
user image
6
 
@PauloCereda Edwin Abbott makes a sandwich.
 
@AlanMunn LOL
 
@DavidCarlisle Some of the required elements are environments though.
 
Apparently I managed to sidestep the issue by figuring out how to convince it to not be in italics mode after all... I was pretty certain that I already tried that. Excuse the spam.
 
@DavidCarlisle But that approach will definitely work for some things.
 
5:26 PM
@AlanMunn hmm ok so your version or somthing like \def\keywordsnotused{\packageerror...} and have \let\keywordsnotused\relax in the keywords environment definition (so no need for actual \if... tests)
 
@PauloCereda would you use emacs on windows?
 
@DavidCarlisle ooh I thought you never get out of emacs
 
No that's vi,
 
@DavidCarlisle Yes, that would work. using \if.. makes the code more transparent, but this is simpler in a way. Thanks.
 
5:28 PM
@AlanMunn and @DavidCarlisle -- That sounds like the right approach for an interactive session, but I gather that I may be the only person on the planet who still works interactively.
 
@barbarabeeton no it's intended for package definitions
 
@barbarabeeton These days most students work with Overleaf, which means not only not interactive, but nonstop mode, so they get output despite dozens of errors...
@barbarabeeton But as @DavidCarlisle says, this is about package/class definitions.
 
@AlanMunn -- Yes. Zillions of errors, most of which are bogus. Sometimes I appreciate being an antique.
 
XOR:
user image
6
 
Just popping to see what's new for me to learn. Off to lunch now...
 
5:40 PM
@PauloCereda What's with that shadow holding the slices? Looks suspicious
 
@PhelypeOleinik ooh a bread plot
 
@PauloCereda Plot twist: it fell with the butter side up!
 
@PhelypeOleinik oooooh
 
user image
3
@PauloCereda ^^^
 
@Rmano oh my
 
6:44 PM
Any suggestions for a github issue label for "bug that wasn't", i.e. effectively user error?
 
@AlanMunn is probably too much Carlislean...
 
@PauloCereda :)
@PauloCereda And too anti-Carlislean.
 
@AlanMunn Quite!
 
@AlanMunn RTFM always gets my vote. :p
 
@AlanMunn, more seriously, does sound a bit rarsh I guess...
 
6:53 PM
@PauloCereda Yeah, that's the problem. I don't want something that make the person look like an idiot, but I'd like to distinguish true bugs from things that weren't.
 
@AlanMunn And is probably not what we want, I guess...
 
@PauloCereda Yes, because that's more for rejected feature requests than bugs that weren't bugs.
 
@AlanMunn indeed...
@Plergux ooh <3
 
@AlanMunn -- How about "documented feature"? (Although "RTFM" is definitely shorter.)
 
or something along the lines?
@barbarabeeton ooh we could use :)
 
7:00 PM
@PauloCereda misinterpretation?
 
@UlrikeFischer do you mean ? Or better yet, !
(what is the markup for tags?)
 
@Skillmon [tag:text-here-without-spaces]
And hi mr. rabbit!
 
@PauloCereda Hi, Mr. Duck! How are you?
 
7:06 PM
@Skillmon I am fine, thanks! And you?
 
I've read that you released a new major version of the cool tool (I still don't use)
 
@PauloCereda Fine, too. My sister visits us this weekend.
@PauloCereda
 
@Skillmon mine is stuck in France which is rather like vi
 
@DavidCarlisle
@Skillmon yaaaay for sis
@DavidCarlisle oh
ooh missing l3pdf thingy
 
7:09 PM
@PauloCereda ?
 
@UlrikeFischer TL mailing list :)
 
@PauloCereda thanks, I answered.
 
@UlrikeFischer ooh you are amazing <3
 
@MarcelKrüger intopdf will need an update, imho it also uses the wrong command names.
 
rather enthusiastic reaction for merely inserting \vspace{} :)
 
7:17 PM
@samcarter_waiting_for_summer who is Mary Me?
 
@UlrikeFischer Maybe the wife of Mini Me?
 
@samcarter_waiting_for_summer did you say yes?
 
7:35 PM
@samcarter_waiting_for_summer I would have said "certainly! but you should know that includes my girlfriend and her husband, my husband and his three girlfriends and their husbands, plus any random paramours we might have at hand at the moment." :p
 
@Skillmon The divide in editor choice seems to big to conquer - a lyx user
 
@samcarter_waiting_for_summer Relatable!
 
@Plergux :) that sounds like one needs a very large moving vehicle
@Skillmon :)
 
@samcarter_waiting_for_summer Well, they come with many benefits, including several moving vehicles. :p
 
@UlrikeFischer I think this is straight and to the point. I might add 'actually' to make it slightly less "you're an idiot" sounding.
I seem to have started a . Thanks for all the suggestions, both serious and humorous. :)
 
7:42 PM
@AlanMunn github labels should be short an concise. If someone is offended by a short and concise summary, that's not your fault.
 
@AlanMunn I've seen in fairly many github repos - seems more or less standard naming
 
@samcarter_waiting_for_summer I remember I once took a quick look at Lyx in the past (before I was this involved in coding in TeX), and found it strange, even back then. But there are certainly quite active users of it...
 
@samcarter_waiting_for_summer So is definitely too long. :) Yes, I think is the simplest.
 
7:56 PM
@DavidCarlisle was already covered by my .
 
@samcarter_waiting_for_summer I have also in circuitikz.
I also have [tag:can't-fix] which is the same but with a bit more guilty feeling
 
@Rmano and there is the enraging [tag:won't-fix] in many projects.
 
@Skillmon I think that come by default in github
 
8:22 PM
@Rmano [tag:can't-be-arsed] :p
@Rmano or you could go the Kevin Wilson way and do :p
 
9:06 PM
[57/74, 01:29/02:38] update: tex.x86_64-cygwin [168k] (55384 -> 57913) ... done
 
9:17 PM
@Plergux ;-P
 
9:31 PM
@UlrikeFischer Oh, I always though I knew all the three users of that package. Which command names do you refer to?
 
@MarcelKrüger I think you still use \pdf_object_now and \pdf_object_last, but we renamed them.
 
@UlrikeFischer note cunning use of plural there to try to deflect some of the blame. It'll never work.
 
@MarcelKrüger the question is on the tl list (Robert Allessi), and he just tumbled over the wrong command names.
 

« first day (3771 days earlier)      last day (1169 days later) »