« first day (3774 days earlier)      last day (53 days later) » 

8:19 AM
Happy Saint David's day!
 
@PauloCereda feast
 
@DavidCarlisle oh no
 
9:18 AM
@Ulrike, is xform generation supported across all engines yet?
 
@AlexG no dvips is still open, and I need to check the box dimension across the engines.
 
@UlrikeFischer ok, thank you!
 
@AlexG I have a remark that I should check pdfbase and atfi-dvips.def for dvips, but I hadn't the time yet (and my postscript isn't the best). If you have some code to offer it would be useful ;-).
 
NEWPAX <3
 
9:34 AM
@UlrikeFischer I think pdfbase is the best I can offer. The dvips version is somewhat tricky, because xform content needs some affine transformations, depending on whether it is used as annotation appearance or referenced in the content stream of a page or a parent xform. Also, the sequence of transformations depend on whether the document is typeset in landscape mode (via doc class option).
I could make an example doc that demonstrates its (pdfbase) usage.
 
9:54 AM
@AlexG that it is tricky is one of the reasons I put it off. Why for example does it depend on landscape? It shouldn't matter if the paper size is 10x20 or 20x10? Also if it depends on if it used as appearance or not, I need to find a good place to add this in the interfaces.
 
@UlrikeFischer There area a few places this comes up: isn't it to do with needing to do work at the PostScript end?
 
@JosephWright where else? And why?
 
@UlrikeFischer Could you put an example somewhere (pdfresources issue?) so we can track this: I doubt I';ll look today but might later in the week
@UlrikeFischer I don't remember right now, but there was some code that needed to know about landscape: was it annotations?
% \subsubsection{Annotations}
%
% \begin{variable}{\g_@@_landscape_bool}
%   There is a bug in \texttt{dvipdfmx}/\XeTeX{} which means annotations do
%   not rotate. As such, we need to know if landscape is active.
%    \begin{macrocode}
\bool_new:N \g_@@_landscape_bool
\cs_if_exist:NT \landscape
  {
    \tl_put_right:Nn \landscape
      { \bool_gset_true:N \g_@@_landscape_bool }
    \tl_put_left:Nn \endlandscape
      { \bool_gset_false:N \g_@@_landscape_bool }
  }
%    \end{macrocode}
% \end{variable}
 
@UlrikeFischer If one sets landscape via class or geometry option, the ps variable /isls in TeXDict is set to true. (By Dvips.)
 
% \begin{macro}{\@@_backend_annotation:nnnn, \@@_backend_annotation_aux:nnnn}
%   Simply pass the raw data through, just dealing with evaluation of dimensions.
%   The only wrinkle is landscape: we have to adjust by hand.
%    \begin{macrocode}
\cs_new_protected:Npn \@@_backend_annotation:nnnn #1#2#3#4
  {
    \bool_if:NTF \g_@@_landscape_bool
      {
         \box_move_up:nn {#2}
           {
             \vbox:n
               {
                 \@@_backend_annotation_aux:nnnn
                   { #2 + #3 } {#1} { 0pt } {#4}
@UlrikeFischer ^^^
 
10:01 AM
@AlexG and which effect has it then?
@JosephWright hm, that is xetex, and if there is really a bug we should check, now that dvipdfmx addresses such bugs rather fast.
 
@UlrikeFischer I'll leave that with you then :)
 
@UlrikeFischer Wrong orientation if no precautions are taken. Please wait for my example.
 
10:14 AM
@UlrikeFischer It looks as though pdfbase needs a fix. Something has changed in the meantime.
 
10:48 AM
@Ulrike, I have to put \leavevmode at the top of my \pbs_pdfxform:nnnnn in the dvips version. Don't know why this is needed now. Maybe because it used picture mode, which I replaced with L3 box moving commands a while ago. (Obviously without testing it thoroughly.)
 
@JosephWright hm, I have disabled your code. You are patching landscape in the backend, and as the backend is loaded before \documentclass the code is lost. I clearly need to think about appearances in this cases as there a not rotated and so squashed if not rectangular, but the xetex code without the landscape fix is in line with the pdftex behaviour for me.
@AlexG getting leavevmode right is a bit of a pain. I always fight with the question is it is needed/useful/harmful in the various cases (as pdflatex errors in a number of cases). Do you know if one can rotate an xform used as appearance? Or do one have to create a second xform with rotated content?
^^^^  the bear is a bit squashed in the checkbox field ;-)
 
@UlrikeFischer :)
 
oh no
 
@Skillmon I see copying files is popular: github.com/latex3/latex3/commit/…
 
@UlrikeFischer pdfbase way of handling this is to tell the intent (appearance or stream referencing) upon xform creation (second arg to \pbs_pdfxform:nnnnn)
 
11:01 AM
@JosephWright ha, fun!
 
@Ulrike How did you create the above example? Using your xform macro?
 
@AlexG yes, but it means I need two xforms if I want to use it in landscape and portrait?
 
@PauloCereda I'm minded to just drop the idea: it's likely to cause issues ...
@PauloCereda I only added them for @yo' ....
 
@JosephWright oh
 
@AlexG yes, along with a number of commands to setup the checkbox structure. I wanted to try to replace at least the hyperref checkbox code which is basically broken.
 
11:04 AM
@UlrikeFischer No, since the landscape setting is done in a per-document fashion (class option) and is immutable.
 
@UlrikeFischer We could kick the 'is it landscape' test to \AtBeginDocument
 
@AlexG you are speaking about dvips? Sorry I was thinking about the landscape environment.
 
(unless the user or a package tinkers with isls PS var.)
 
@JosephWright interesting timing indeed :)
 
@UlrikeFischer No, the landscape env does no harm.
 
11:06 AM
@JosephWright As I wrote I think it is for xelatex no longer needed. Or at least it will need new tests first, to really check this.
@AlexG Didn't you see the bear?? I would call this harm ;-)
 
@UlrikeFischer Ah, right: log an issue anyway?
 
@UlrikeFischer Oh my, I need to check and build my own example. I am not as fast as you in putting things together.
 
@AlexG but you are also awesome! <3
(")>/
 
@AlexG I'm only fast because I spent yesterday evening updating the checkfield code to the various changes in pdfmanagement and so have a few testfiles around.
 
@PauloCereda <3 Thank you for your cheering me up!
@UlrikeFischer Phew, I was afraid you were supernatural.
 
11:12 AM
@AlexG but I still don't quite understand why it is there. It feels like a work around if dvips doesn't get the right papersize by other means (the \special}, but nowadays this it normally there.
 
@UlrikeFischer We should look again at that
 
@UlrikeFischer Yes, this isls thing looks indeed like a relic. But I don't think it has any relation to landscape setting via environment.
 
@AlexG yes, this are two different questions. But your code contains a number of lines related to this isls and before moving this into the pdfmanagement code it would make sense to find out, if one can remove it safely (or what is required to get rid of it). So what it is doing in texdict?
 
@UlrikeFischer I must admit that I don't understand the PS code in the TeXDict. Whom to ask? T. Rokicky?
But since isls is set once per document (at the beginning), it can be handled upon xform generation. The different ways of xform generation for appearances and stream referencing is more of a concern, I think.
 
11:29 AM
@AlexG how is isls changed if landscape is used? And have you and example for the appearance/stream problem?
 
It shouldn't be changed by setting landscape environment. I am just starting to work on my test example.
 
@UlrikeFischer Do you have an example for the XeTeX thing?
 
@JosephWright the moving? I will try make one which doesn't require all the checkbox stuff.
 
@UlrikeFischer Cool
 
11:47 AM
@UlrikeFischer Stream usage is straightforward. First, a lscape pkg example. Same output with pdftex and dvips:
\documentclass{article}
\usepackage{pdfbase}
\usepackage{graphicx}
\usepackage{pdflscape}

\begin{document}
\sbox0{\includegraphics[width=1in]{example-image}}%
\ExplSyntaxOn
   \leavevmode\pbs_pdfxform:nnnnn{}{0}{}{}{0}
   Referenced~xform:\pbs_pdfrefxform:n{\pbs_pdflastxform:}
\ExplSyntaxOff

\begin{landscape}
\ExplSyntaxOn
   Referenced~xform:\pbs_pdfrefxform:n{\pbs_pdflastxform:}
\ExplSyntaxOff
\end{landscape}
\end{document}
 
@JosephWright this here should work as test: for me I get the pdflatex result if the boolean is false on a current texlive:
 
This is landscape as class option (isls-->true):
 
%\RequirePackage{pdfmanagement-testphase}
%\DeclareDocumentMetadata{uncompress}

\documentclass{article}
\usepackage{pdflscape}

\begin{document}
\ExplSyntaxOn
\cs_set_eq:NN \pdfannot_box:nnnn \__pdf_backend_annotation:nnnn

\bool_gset_false:N\g__pdf_landscape_bool

xxx \pdfannot_box:nnnn
 {1cm} {2cm} {0cm}
  {/Subtype/Link/A<</Type /Action /S /URI /URI (blub)>>/Border[0~0~1]/C[1~0~0]}{XXX}
%
\bool_gset_true:N\g__pdf_landscape_bool

\quad
xxx \pdfannot_box:nnnn{1cm} {2cm} {0cm}
  {/Subtype/Link/A<</Type /Action /S /URI /URI (blub)>>/Border[0~0~1]/C[1~0~0]}{XXX}
 
\documentclass[landscape]{article}
\usepackage{pdfbase}
\usepackage{graphicx}

\begin{document}
\sbox0{\includegraphics[width=1in]{example-image}}%
\ExplSyntaxOn
   \leavevmode\pbs_pdfxform:nnnnn{}{0}{}{}{0}
   Referenced~xform:\pbs_pdfrefxform:n{\pbs_pdflastxform:}
\ExplSyntaxOff
\end{document}
 
@AlexG @JosephWright but I have to make lunch now ...
 
11:54 AM
Me2.
 
@UlrikeFischer oh... no?
It's an automatic trigger. :)
 
@UlrikeFischer OK, so we drop that switch and 'tough' for anyone it doesn't work for
 
12:42 PM
@UlrikeFischer Of course, inside landscape environment, the annotation's width and height args need to be swapped.
@Ulrike, please don't tweak the original \pdfxform behaviour in the l3 interface (introducing some automatic width<-> height swapping based on landscape environment).
 
12:57 PM
@AlexG I didn't plan to do this, this looks quite wrong. landscape is a high-level environment, and backend code shouldn't have branches related to this. What one could do is to add a boolean "swap", which can be set in environments.
 
@UlrikeFischer That's good. Thank you.
 
1:26 PM
baa
 
yo'
@PauloCereda boo from another full lockdown.
 
@yo' ooh
@yo' /hugs
 
yo'
@PauloCereda thanks, but it might not be the safest thing to do, I don't feel well again ...
 
@yo' ooh no!
 
@yo' oh no
@Skillmon quack
 
yo'
1:33 PM
Somehow guys you got the number of o's all mangled up... :)
Anyway, need to do some work before my body decides I'm too weak to be at the PC...
 
@yo' mr. rabbit is trying to counter-ooh my ooh's :)
 
yo'
@PauloCereda o{1,2}h
 
@yo' ooh
 
1:52 PM
@PauloCereda I had to help out since you took so long.
@PauloCereda counter-counter-ooh and counter-ooh?
 
@Skillmon ooh :)
@Skillmon <3
 
@PauloCereda ad infinitum?
 
@Skillmon here we go :)
 
2:10 PM
@yo' Get well soon!
user image
3
 
yo'
@samcarter_waiting_for_summer thanks! Got AG test booked for tomorrow to check whether it's just be being tired after a weekend in a cottage or whether it's something... so let's hope
 
@samcarter_waiting_for_summer lovely! I used to remember how to say nurse in German, but now...
something something schwester
I know there's sister somewhere
 
@PauloCereda You're totally right, it is "Krakenschwester"
 
@samcarter_waiting_for_summer ooh
 
@yo' I'll keep my fingers crossed!
 
2:18 PM
@samcarter_waiting_for_summer I think there is an n missing, but Krakenschwester is funny as well.
 
@Skillmon ups :)
user image
6
 
@samcarter_waiting_for_summer oh my
 
@samcarter_waiting_for_summer ooh, a Krakenkrankenschwester
 
How would you say in German, "Sister Kraken Nurse"?
Krakenkrankenschwesterschwester?
3
 
@PauloCereda Only one "Schwester" and without an upper case K for "Kranken" (essentially, see my message)
 
2:22 PM
@Skillmon oh
Fixed :)
 
So how would “Sister Kraken Nurse's Kraken Nurse Sister” be?
Krakenkrankenschwesterschwesterkrakenkrankenschwesterschwester?
2
 
@PhelypeOleinik OH MY YES
 
@PhelypeOleinik XD
 
@Plergux /poke
 
2:47 PM
@PauloCereda /jab
:p
 
@Plergux oh
 
3:19 PM
@samcarter_waiting_for_summer How on earth did you find this image? Looking for Krakenschwester on Google Images didn't help...
Found this on the 9th page: startpage.com/av/…
 
@AlexG searched for "octopus nurse" on duckduckgo and this was the 5th hit
@AlexG that look scary!
 
@samcarter_waiting_for_summer I used google via startpage. Yes, indeed. I prefer yours, definitely.
 
@AlexG All hail to Cthulhu?
 
@AlexG The difficult part was to decide which colour to pick. The purple one also looked nice.
 
@Skillmon Oh, a knowledge gap.
 
3:42 PM
@UlrikeFischer A pdfbase example with the two xform versions (for content stream referencing and for annotation appearance) used as appearance streams in a landscape document. Oh my, just saw that landscape as class option and as pkg geometry option need different handling. Currently, pdfbase supports [landscape]{geometry}:
\documentclass{article}
\usepackage[landscape]{geometry}

\usepackage{pdfbase}
\usepackage{graphicx}

\begin{document}

\sbox0{\includegraphics[width=1in,height=0.75in]{example-image}}%
\ExplSyntaxOn
\leavevmode\pbs_pdfxform:nnnnn{}{0}{}{}{0} % for reffing in streams, wrong here (appearance use)
Button:\pbs_pdfannot:nnnn{1in}{0.75in}{0pt}{
  /Subtype/Widget
  /Ft/Btn/Ff~65537/T~(myButtonWidget0)
  /AP<</N~\pbs_pdflastxform:>>
}
\ExplSyntaxOff

\vspace{2in}

\sbox0{\includegraphics[width=1in,height=0.75in]{example-image}}%
 
3:57 PM
Another day in the kernel, @JosephWright looks worried ^^
 
@AlexG hm, I see no appearance at all in acrobat reader.
 
I think the changes to the option handling (as they are done according to ltnews 33) will lead to quite a few packages having to be adapted. Changing the contents of the unused options list to not include the value will trip them... (including my small package, which will be entirely rewritten once I had the time to look at the new possibilities)
 
@UlrikeFischer Did you run dvips? Also, -Ppdf must not be used with dvips.
 
@DavidCarlisle, @UlrikeFischer Seen the TL mailing list post re. xetex.def?
 
@AlexG yes.
@JosephWright not yet, will look.
 
4:04 PM
@UlrikeFischer Works for me in Evince & Okular. Yet need to check with AR
 
@Skillmon well trouble is if you have {} in a value the \if@in test gives a low level csname error which means that the idea of saving the original raw options to habd on to a keyval package doesn't really work at all, you still have arbitrary restrictions on values. But latex-dev is -dev for a reason This is a scary area and we have backed it out and not changed it more than once so comments welcome. (gh better than here for this)
@JosephWright yes but can't look now
 
@DavidCarlisle sure, I know why it's in -dev... I just read the news and this came to mind. The new route to just store everything up to = in the unused option list is reasonable, just that this will require all packages which deal with the unused option list to be adjusted (or did you also change \@removeelement?).
 
@UlrikeFischer Weird. AR doesn't show the buttons.
 
@Skillmon if all packages have to be told that's a bug. but can't look now sorry
 
@DavidCarlisle haven't yet looked into the real implementation, just read the news, which gave this impression. I might be wrong here.
 
4:23 PM
@yo' -- Please get well!
 
@UlrikeFischer Sorry, a typo in the code: s/\\Ft/\\FT/. The corrected example for ps2pdf
\documentclass{article}
\usepackage[landscape]{geometry}

\usepackage{pdfbase}
\usepackage{graphicx}

\begin{document}

\sbox0{\includegraphics[width=1in,height=0.75in]{example-image}}%
\ExplSyntaxOn
\leavevmode\pbs_pdfxform:nnnnn{}{0}{}{}{0} % for reffing in streams, wrong here (appearance use)
Button:\pbs_pdfannot:nnnn{1in}{0.75in}{0pt}{
  /Subtype/Widget
  /FT/Btn/Ff~65537/T~(myButtonWidget0)
  /AP<</N~\pbs_pdflastxform:>>
}
\ExplSyntaxOff

\vspace{2in}

\sbox0{\includegraphics[width=1in,height=0.75in]{example-image}}%
 
@PauloCereda ohh :)
 
4:41 PM
user image
5
Fubá says hi. :)
 
@PauloCereda -- More like Fubá say g'night.
 
@barbarabeeton :)
 
@DavidCarlisle Another arydshln conflict. tex.stackexchange.com/q/585428/2693
 
@PauloCereda Fubá reminds me of this advertisement: youtube.com/watch?v=OC-LKpI2k3A
 
@AlexG if I remove the landscape, the first button is mirrored/upsidedown. Why?
 
4:54 PM
@samcarter_waiting_for_summer ooh
 
@UlrikeFischer This is the dvips weirdnes that pdfbase tries to handle: xform for annotation appearance vs. xform for content stream referencing. Transformations are needed in the former case, (as much for portrait as for landscape; in landscape, needed transformations are a bit more complex, mirroring is not enough)
 
@AlanMunn not my fault or I don't care or blame @UlrikeFischer
3
 
@DavidCarlisle :) that such a package is still needed (and not available more generally) has always puzzled me.
 
@AlanMunn yes we probably should add dotted rules to array package they seem weirdly popular.
 
5:09 PM
@JosephWright I don't think that the color problem is due to the stacks, but the parsing is wrong, an convert@color@ {NavyBlue} ends up in the stream.
 
@UlrikeFischer True: I've obviously messed something up
 
@JosephWright we had such an expansion problem already. I wonder if it was corrected in the sources?
 
@UlrikeFischer Ah, could be it
 
@JosephWright the discussion is here, but the solution seems not to change something chat.stackexchange.com/transcript/message/56811366#56811366
 
5:39 PM
@UlrikeFischer Need to alter how named colours are handled: sigh
 
@JosephWright this named colors are quite a pain ;-). With xcolor it works.
 
5:58 PM
@JosephWright there seems to be a reason for @DavidCarlisle being monochrome... he knew
2
 
Quoting a quote: "EM cannot process LaTeX files with macros" :-O (tex.stackexchange.com/questions/585432/…)
I do not know who or what EM is, but that's a funny thing to say...
 
ooh Peter's mum
 
@Rmano Editorial Manager (from Elsevier or Springer, I don't remember which exactly). If I remember correctly, it can, but it's a really odd (and very limited) environment, so an user package is unlikely to work
 
6:13 PM
@PauloCereda Is this your tip of the day?
 
@samcarter_waiting_for_summer ooh I could make this into a series
 
@PauloCereda yessss
 
@samcarter_waiting_for_summer hmmmmm
 
@samcarter_waiting_for_summer By the way, thanks for the points on that answer :)
 
@PhelypeOleinik Given that everything is a macro in LaTeX --- some macro they should be able to process ;-). Anyway, removing the possibility of user macros and still calling it LaTeX is a bit ...misleading, no?
 
6:17 PM
A minister, a priest and a rabbit walk into a bar. The bartender looks at the rabbit and says, "Sorry, there's been a clerical error."
7
 
@AlanMunn ooh
 
@Rmano Indeed! Who wrote that likely didn't know exactly what it was all about and used some words found on the internet. Google LaTeX and you inevitably find scary stuff called macros, so better avoid these :)
 
@PhelypeOleinik You're welcome! With all the comments etc it is really a nice to read and interesting answer
 
@PhelypeOleinik Well, I think that they do not mean LaTeX macros at all. If you look at page 8 of the document found by @samcarter_waiting_for_summer, they allow even personal .sty, so I think that they mean something at the OS level. What they mean, really, is a mystery...
 
@Rmano ^^^ I think this tells everything one needs to know about the EM thingy
 
6:22 PM
@Rmano I think a couple years ago I submitted a paper to this EM thing, and was in doubt if they were really building my LaTeX document in their server (I think it's Springer, because Elsevier takes the PDF as is). I remember I used some \IfFileExists trick with a local file to find that out, so macros definitely work :)
@samcarter_waiting_for_summer Glad you liked :D
 
@Rmano perhaps they meant Macros el Evangelista disléxico :)
That was awful, sorry
 
@PauloCereda Or maybe it was a typo and they meant macos (MacOS). That would be a reasonable restriction :)
 
@PhelypeOleinik my wild guess is they do not allow for --shell-escape.
 
@PhelypeOleinik oh no, macos :D
 
@Rmano That's probably true (and also reasonable)
@PauloCereda :)
 
6:26 PM
@PhelypeOleinik Yes. But if it is this, it is quite awfully stated...
 
@Rmano To say the least :)
 
@PhelypeOleinik grrr
 
@AlanMunn Sorry, I couldn't miss the joke :)
 
@PhelypeOleinik :)
 
@PhelypeOleinik nice challenge: How do one adjust hook code in AtBeginDvi that is added in AtBeginDocument?
 
6:34 PM
@UlrikeFischer You mean like \AtBeginDocument{\AtBeginDvi{<sutff>}}? Do you have an example?
 
@PhelypeOleinik geometry does this (I was trying to test what happens if I remove its dvips code).
 
@UlrikeFischer Hm, I'll check
 
@PhelypeOleinik With the new hook management, it should naturally simply use \AtBeginDvi and setup rules, if it wants to be late or early or whatever.
 
@UlrikeFischer You mean the line \AtBeginDvi{\special{papersize=\the\paperwidth,\the\paperheight}}%?
@UlrikeFischer Yes, it looks like it's trying to delay the \special in case some other package that sets the page dimensions is loaded after geometry
 
@PhelypeOleinik I was more interested in the \AtBeginDvi{\special{! /landplus90 true store}}, but it is the same problem.
@PhelypeOleinik yes, but \AtBeginDvi would work in any case, so why hiding it in AtBeginDocument too?
 
6:42 PM
@UlrikeFischer typearea seems to do roughly the same (not sure how much that's used for KOMA classes)
 
@PhelypeOleinik this paper setting stuff is everywhere (but I removed it from the new hyperref driver so one less in the crowd).
 
@UlrikeFischer Suspicious occurrences of landplus90:
latex/pecha/pecha.cls
latex/geometry/geometry.sty
latex/koma-script/typearea.sty
latex/powerdot/powerdot.cls
latex/uwmslide/uwmslide.cls
@UlrikeFischer I didn't find that in hyperref though...
 
6:59 PM
@UlrikeFischer they worked brilliantly until you broke them?
 
@DavidCarlisle ooh
 
@UlrikeFischer Did you see this?
%    If |dvipdfm| is specified and \textsf{atbegshi} package in
%    `oberdiek' bundle is loaded, \cs{AtBeginShipoutFirst} is used
%    instead of \cs{AtBeginDvi} for compatibility with \textsf{hyperref}
%    and |dvipdfm| program.
 
@PhelypeOleinik Want push access to a github repo?....
 
7:16 PM
@PhelypeOleinik I think hyperref doesn't has this landplus code, but it has code to set the pdf page size, this one I removed.
@PhelypeOleinik where is it from?
 
@UlrikeFischer geometry
@UlrikeFischer the \ifHy@setpagesize code?
 
@AlanMunn good joke. Now
 
@DavidCarlisle yes, at least the part in the driver, hyperref.sty itself is unchanged (but if we are daring we could set the boolean to false and look if someone complains)
 
@UlrikeFischer I don't think we can do that until there is some approved way to set the page size though? Been telling people for decades that you need geometry or hyperref or graphicx before latex sets the page size
 
@PauloCereda I was wondering, why doesn't arara stick to the first $n$ lines and the last $m$ lines (both extensible for consecutive statements), instead for searching the entire file? The latter (and current) behaviour just serves for obfuscation and doesn't add any real value to the rules (one could argue that such possibility of obfuscation is a security risk in some sense)
 
7:27 PM
@DavidCarlisle yes, but the backend could do it now. I mean unlike the color/graphics driver it is loaded always.
 
@Skillmon one could use the --header flag which make arara fetch only directives specified in the header. :)
 
@UlrikeFischer yes sure but unless I missed something that means taking the default or loading expl3
 
@DavidCarlisle sorry? what default?
 
@PauloCereda ooh, sorry then (I'm not really familiar with arara, still, I'm just too accustomed to GNU make...
 
@Skillmon no worries. <3
 
7:30 PM
@PauloCereda but still, imho, it would be sensible to switch to --header being the default in that case, and allowing a --full option.
 
@UlrikeFischer the default back end (so dvips not dvipdfm or dvisvg)
 
@DavidCarlisle but if you want to use dvisvg or dvipdfm you have to tell it l3backend in any case if don't want problems. One l3backend file is always loaded.
 
@UlrikeFischer yep, so we should make all of geometry/hyperref/graphicx use the l3backend but we're not quite there yet are we? perhaps it's OK simply to delete ...
 
@DavidCarlisle I don't think we have to make the other package to do something. It would be enough if l3backend would set the pagesize too, then there would be one method which is reliably always there and the other could slowly stop to do it too.
 
@UlrikeFischer yes that's what I mean, drop the code from them and rely on l3backend, but it would also be good if \usepackage[dvipdfm]{graphicx} forced l3backend to load the matchig code
 
7:39 PM
@DavidCarlisle but this problem is already there (and I sometimes wonder why they aren't not more question about it). Beside this: with the pdfmanagement the backend is always loaded before graphics, so it is more a question of graphicx detecting which backend has been set in \DeclareDocumentMetadata.
 
@DavidCarlisle Access doesn't mean I do push, so sure :)
 
@UlrikeFischer well yes but none of my documents have got \DeclareDocumentMetadata. yet:-) (but seriously we should fix this)
 
@UlrikeFischer geometry.dtx. Has slightly more info than the .sty, but nothing too helpful
 
@DavidCarlisle well it is one goal of the new setup to have one place to declare the backend, and to avoid \usepackage[dvisvg]{color}\usepackage[dvipdfm]{graphics} \usepackage[texttures]{hyperref} ;-)
 
@UlrikeFischer it's my fault it's even possible to do that :(
 
7:46 PM
@DavidCarlisle Which repo is it?
 
@DavidCarlisle Oh, happy new year
 
@DavidCarlisle I don't see how you could avoid it, I mean every package is free to define backend options (including slightly different name like dvipdf, dvipdfm, dvipdfmx ...).
 
@DavidCarlisle Of course you maintain it :)
 
@UlrikeFischer yes but three packages with backend option code by the same author.... I almost got it right with color and graphicx sharing the same .def file but then messed up with them being able to specify the options independently
@PhelypeOleinik one of the ones I aquired for "emergency fixes because luatex changed"
 
7:54 PM
@DavidCarlisle are you really maintaining it or is it still only for emergencies?
 
@UlrikeFischer I'm still only "standing in" but I think I have unrestricted ctan upload rights....
 
@PhelypeOleinik I thought that too when I accepted push access to hyperref ...
 
@UlrikeFischer or I picked up one or two packages under oberdiek...
 
@UlrikeFischer TrapHub
 
8:41 PM
@Skillmon one can make it a default behavior via configuration file... changing the default behaviour is a breaking change.
@PhelypeOleinik :)
 

« first day (3774 days earlier)      last day (53 days later) »