« first day (5120 days earlier)      last day (32 days later) » 

cfr
cfr
02:32
@UlrikeFischer that's why I said I assumed it was expected. it works fine if I test the format version first. thank you.
 
4 hours later…
07:00
@DavidCarlisle, @barbarabeeton It's just the design choice - scrolls fine ... I'm no web designer!
07:40
@JosephWright true but with no scrollbar there is no indication scroll is needed, and with some interfaces scrolling may be harder
@JosephWright oh, accessibility problems...
08:32
@DavidCarlisle I meant that normally you say 'design decisions are made as a whole - it's not really the place of a user to change them' - but I'll see what I can do
08:44
@JosephWright ah agreed, although I thought you meant it was your design:-) I hadn't noticed myself as I typically use a scroll gesture, but I think most sites do keep the scrollbar, if only to show that scrolling is needed, eg this one where the bar auto hides, and is impossibly thin to actually use, but does appear so you get an indication of where you are, and it would make @barbarabeeton happy:-)
@DavidCarlisle Try now
@DavidCarlisle wednesday quack
@PauloCereda Quack
@DavidCarlisle Took a couple of attempts to find the right place to alter the CSS
@JosephWright @barbarabeeton will be happy:-)
@JosephWright ooh a duck
08:55
@PauloCereda wednesday breakfast
@PauloCereda double helping
@DavidCarlisle oh no
@DavidCarlisle oh no
@PauloCereda Matrix
I think the era of Skype is over ;)
@JosephWright do you think I should let @PauloCereda know my super secret matrix account name so he can contact me there?
@DavidCarlisle Sounds like a plan to me :)
@DavidCarlisle ooh
09:25
@Skillmon It seems that it is somehow related to a warning message that (instead of issuing a warning) raised an error. I wonder why that is
\__kernel_tl_set:Nx
\tl_set:Nn #1#2->\__kernel_tl_set:Nx
                                     #1{\__kernel_exp_not:w {#2}}
<argument> \tl_set:Nn \l__msg_current_class_tl {warning}
                                             \cs_set_protected:Npe \__msg_use_c

\use_ii:nn #1#2->#2

\__msg_use:nnnnnnn ...nnnn {msg}{unknown}{#2}{#3}}
                                                  \cs_if_exist_use:N \condit...
<argument> \msg_warning:nn {zchinr}{reference-unknown}
                                           \textbf {?}
How should I read above error, if it says `./issue.ptc:10: Missing number, treated as zero.
<to be read again>` before this information?
@JasperHabicht show more context.
@DavidCarlisle how to do that if he doesn't look into skype ;-(
@JasperHabicht Looks like a non-expandable warning in an expansion context
@JasperHabicht Really do need a full example to help more
09:42
@JosephWright Aha! This can really be the problem ... But why would this not raise an error with XeLaTeX ... Thanks, this is good hint!
@JasperHabicht If there's a different code path - but without an example, like I said, can't really say more
I know, this is very little context. I am sorry for that. I tried to come up with something that can make this better understandable, but it is very difficult to do this. I'll go on with this idea and check again. Thanks!
09:57
Hah, I found it! Oh my ...
Thanks anyways! And sorry for the noise!
10:15
@JasperHabicht Do tell :)
10:28
@JosephWright The issue was that I have an \int_compare:nNn to which I fed something different that a number if the reference was not found. Of course, this would raise an error. It seems that it did not raise an error before only because I already compiled everything a few times so that there never was an unknown reference.
@JasperHabicht Ah, that would do it :)
But since this was covered deepely in nested code structures, it was really not easy to find
 
6 hours later…
16:02
@cfr The polyglossia Linux Libertine issue is very strange. Seems to only happen when the class is similar to book (tested with book, memoir, scrbook but does not seem happen with report)
cfr
cfr
16:38
@daleif what you describe definitely sounds very strange. you get problems with upstream TL? what does the debian font package provide?
\documentclass{beamer}
\usepackage{fontspec}
%\usepackage{unicode-math}
\begin{document}
\begin{frame}{Bullets}
\begin{itemize}
\item An item
\begin{itemize}
\item A subitem
\end{itemize}
\end{itemize}
\end{frame}
\end{document}
Why does unicode-math make the bullet symbols too large?
% !TeX TS-program = lualatex
\documentclass{beamer}
\usepackage{fontspec}
\usepackage{unicode-math}
\begin{document}
\begin{frame}
\frametitle{Bullets}
$\blacktriangleright$
\end{frame}
\end{document}
@AlanMunn Different font for \blacktriangleright?
% !TeX TS-program = lualatex
\documentclass{article}
\usepackage{amssymb}
\usepackage{fontspec}
\usepackage{unicode-math}
\begin{document}
text $\blacktriangleright$
\end{document}
@AlanMunn You'd get the same in article
@cfr seems to just be font files in a system location. In the article case I haven't checked where exactly it takes the font from just that Linux Libertine O works. And it also depends on polyglossia being used or not.
16:54
@JosephWright -- Oh, yes, I'm happy now. Thank you! (My "default laptop" is flaky, and has a nasty habit of the screen going dark. The non-alphabetic functions o the backup operate differently, and inedge cases, may not work at all. Very disruptive.)
cfr
cfr
17:09
@daleif what does \chapter do that makes a difference, though?
17:25
@JosephWright Nice, some notes (at least a link) to CJK should be added somewhere
In my recent document class, I just changed:
\sys_if_engine_luatex:T {
  \RequirePackage{luatexja-fontspec}
  \setmainjfont[ItalicFont={Noto ~ Serif ~ CJK ~ SC}]
    {Noto ~ Serif ~ CJK ~ SC}
  \setsansjfont[ItalicFont={Noto ~ Serif ~ CJK ~ SC}]
    {Noto ~ Sans ~ CJK ~ SC}

  % prevent German quotation marks from being spaced
  \ltjsetparameter{
    alxspmode = {"201E, preonly},
    alxspmode = {"201C, postonly}
  }
}

\sys_if_engine_xetex:T {
  \RequirePackage{xeCJK}
  \setCJKmainfont[ItalicFont={Noto ~ Serif ~ CJK ~ SC}]
    {Noto ~ Serif ~ CJK ~ SC}
So, it is really more or less chaning xeCJK to luatexja-fontspec, but of course certain things have to be adjusted manually, due to the different ways that XeLaTeX and LuaLaTeX work
17:51
@cfr I could not find any thing on \chapter and report works but not book, memoir, scrbook...
 
1 hour later…
cfr
cfr
19:02
@daleif I thought you meant the error occurred only after \chapter but you mean only if the class provides \chapter, right? so this errors:
\documentclass{book}
\usepackage{polyglossia} %
\setmainlanguage[variant=british]{english}
\setmainfont{Minion Pro}
\begin{document}
prawf
\end{document}
but change book to article and all is hunky dory?
@daleif ironically, it works fine with xetex ....
19:16
Something like that, but report worked. And if polyglossia is replaced by fontspec all works with all classes. Which is why I do not understand
@samcarter Annoying. :)
@daleif Does bold also work if you load it like that? After the last days of discussions, it seems that loading like \setmainfont{Minion Pro} is not so reliable.
@mickep not at pc. I'll got boot up
@daleif Even if it does, it seems safer to load by filename.
@daleif polyglossia forces all fonts to default to Harfbuzz mode in luatex, so check if you can get the error using Render=Harfbuzz in luatex (not actually read all the thread to see what the error is:-)
cfr
cfr
19:29
@DavidCarlisle also errors.
@cfr as I guessed:-)
@cfr at home I don't get the loop warning with otfinfo -i, not 100% use that I have the full installation here at hone.
cfr
cfr
@daleif ?? the warning is from fontconfig.
@cfr right, I also get the looping line at home
@DavidCarlisle you mean \setmainfont{Linux Libertine O}[Renderer=Harfbuzz], that still works with just fontspec and article but not with book.
So seems to be related, but don't understand why article works.
\documentclass{book}
%\documentclass{article}
% \usepackage{polyglossia} %
\usepackage{fontspec}
\setmainfont{Linux Libertine O}[Renderer=Harfbuzz] %
\begin{document}
test
\end{document}
luaotfload | db : Reload initiated (formats: otf,ttf,ttc); reason: Font "Linux" not found.
! Font \TU/LinuxLibertineO(0)/m/it/10=name:Linux not loadable: metric data not
found or bad.
<to be read again>
L
l.8 \end{document}
Why is it only reading Linux.
Was thinking that it was Expl mode removing spaces, but then the error ought to be LinuxLiberTineO.
19:45
@daleif First compilation I tried says luaotfload | db : Reload initiated (formats: otf,ttf,ttc); reason: Font "Linux Libertine O" not found." For the second it is just luaotfload | db : Reload initiated (formats: otf,ttf,ttc); reason: Font "Linux" not found..
@daleif With this very example.
@mickep interesting. Anyone remember how to reset the luaotfload cache?
@mickep I do not seem to be able to replicate that. Even after erasing the luaotfload cache, I only get the "Linux" error.
@daleif Weird. Loading by name works again. Changing Linux Libertine O into Linux Libertine R (for example) results in a different error.
@mickep But do you have two libertine fonts installed (I haven't checked to see if I have the TL one as well). If I replace O by R, then it complains about the full name. Not just Linux. It this something about the family or subfamily name again?
@daleif I only have the texlive one installed. I do not understand how the \setmainfont works...
But does Linux Libertine R even exist among the font names in the files? otfinfo -i /usr/share/fonts/opentype/linux-libertine/LinLibertine_R.otf does list Linux Libertine O as the family and full name.
I seem to have both (TL and ubuntu)
20:00
@daleif No, that was just an example of something non-existing. So, it finds something (even though the message indicates it is not found).
@mickep Right, so if the name is not found it lists the full name. Here it seems to find something, do something strange, and gets confused.
@daleif Since it works without HarfBuzz, the problem maybe lies there...
@daleif it works with the dev branch of luaotfload, and I think it is related to this issue github.com/latex3/fontspec/issues/486 but I have no time yet to check which font call in book is triggering it.
20:49
@daleif the difference between book and article is that book calls \slshape in the header.
cfr
cfr
@daleif Yes, because you have the font. @mickep gets an error the first time because (I assume) Linux Libertine O is not installed on the system; you installed it, so there is no error. that's because fontspec has no problem so long as Harfbuzz isn't used. the font family is set up correctly with italic etc. etc. only later do you get an error in that case.
@UlrikeFischer would that be renderer specific?
@cfr Do you by "on the system" mean outside of TL?
cfr
cfr
@mickep well, it depends on your setup. if you configured the system to use fonts in TL, they should be included. if not, I'm not sure ...
@cfr I did not set anything up, really. Just installed TL on a debian system. Maybe added something to $PATH.
cfr
cfr
21:05
@mickep then I guess maybe you need to do that. there's a drop-in fontconfig file you can use to do it, if you want.
there's an issue somewhere on github which might be related. I can't find that one, but github.com/latex3/luaotfload/issues/285 looks likely to me. but @UlrikeFischer surely knows better.
@cfr I prefer not to do any tweaks. I just tested the document snippet from @daleif, it is not that I use this. I installed the libertine fonts with "tlmgr install libertine". I thought that would do everything needed.
cfr
cfr
@mickep I'm not sure. I try not to dabble mostly in 8bit when I need black magic.
(I just wanted to test @daleif's snippet and report back)
cfr
cfr
@mickep it is not exactly a tweak. the file is in TL. if you installed your distro package of TL, it would likely be installed for you. so I don't see it as any different from adjusting PATH and the search for info. it's just part of setup if you use upstream. if you alter the contents of the file (as I have), that's a tweak. but just installing it ... no. /usr/local/texlive/YYYY/texmf-var/fonts/conf/texlive-fontconfig.conf
@cfr I have that file.
cfr
cfr
21:15
but, if you do install it, you should do it right ;).
@mickep of course you do. but you have to drop it into the right place.
-rw-r--r-- 1 root root 460 Hyd  17  2012 /etc/fonts/conf.avail/09-texlive-fonts.conf
lrwxrwxrwx 1 root root  35 Hyd  13  2017 /etc/fonts/conf.d/09-texlive-fonts.conf -> ../conf.avail/09-texlive-fonts.conf
Since I did tweak it, the 2012 copy still works in 2024 ....
@mickep but fontconfig is very black magic, so I have to be very sure before I meddle in those directories.
@cfr Hm, OK. I think I leave it as it is. Thanks, though!
cfr
cfr
21:40
@mickep this particular file would be safe enough. all it does is add TL directories so fontconfig can find fonts in those as well as the system ones. it doesn't actually change your font configuration. installing it into /etc/conf.d wouldn't even add the directories. files there are ignored. they're only used via sym links in /etc/conf.avail. (hence the circuitous setup above.) but, if you don't miss them, there's no reason to do it.
can l3build copy supplementary files from the system texmf prior to sandboxing tests, say?

« first day (5120 days earlier)      last day (32 days later) »