« first day (3044 days earlier)      last day (2109 days later) » 

00:08
@egreg As wrong as a pineapple on a pizza? ;-)
00:34
@UlrikeFischer hmmm
 
8 hours later…
09:04
@UlrikeFischer @JosephWright spot the difference:
$ tex
This is TeX, Version 3.14159265 (TeX Live 2019) (preloaded format=tex)
**
! End of file on the terminal... why?

$ luatex

$
@DavidCarlisle ooh
@DavidCarlisle I get a normal ** prompt.
09:21
@FaheemMitha I get nothing it just aborts with no error or log, updating cygwin first, before querying it.
@PauloCereda you haven't been working on texdoc have you?
@DavidCarlisle uh-oh... what happened?
@DavidCarlisle ;-). oberdiek fails a lot too, as far as I could see only (but I didn't check every failure) because of \expanded. But I think the testsuites can't work in tl18 and tl19. So how (and when) do we switch to tl19?
@JosephWright vvv
> Periodic Table Turns 150 Years Old
@PauloCereda nothing I started by noticing texdoc wasn't giving doc, before noticing that texlua wasn't working at all
@DavidCarlisle oh no
09:24
@UlrikeFischer see discussion with @JosephWright yesterday, depends a bit on how we want to handle travis
@DavidCarlisle I'm sorry, I don't follow. Context, please.
@PauloCereda and I noticed something in the texdoc sources that made me suspect you...
@DavidCarlisle uh-oh
@FaheemMitha with texlive 2019 if I type texdoc or luatex or texlua etc, then the command just exits, no error no log just nothing.
09:27
@DavidCarlisle are you sure you want to answer to @FaheemMitha?
@UlrikeFischer Heh. I was about to ask that.
@DavidCarlisle I think you meant to address that to someone else.
@FaheemMitha no you asked for the context of my previous comment, so I answered giving more context?
@DavidCarlisle but it's Lua-related stuff, not ConTeXt... /ba dum tss
@UlrikeFischer @PauloCereda meanwhile texlive 2019 luatex now works (probably everything else broke:-) I did a system update to cywin 3.0 .....
@DavidCarlisle woo things broke
09:31
@DavidCarlisle You addressed that one to me. Which I think you did not mean to. And the comment you linked to was "Mercurial has sequential numbering."
Which is clearly unrelated.
@FaheemMitha ah:-)
@PauloCereda tail -1 /usr/local/texlive/2019/texmf-dist/scripts/texdoc/texdoc.tlu :(
@DavidCarlisle LOL
09:45
@DavidCarlisle People do still use Mercurial. Though the project has adopted at least one dreadful policy, which makes me worry about their future.
Someone, please ask me what it is.
@FaheemMitha what is it
@DavidCarlisle no who is it
@DavidCarlisle I'm glad you asked. They autoclose bugs (using a bot) that are sufficiently old, with the status "RESOLVED ARCHIVED". But without actually fixing the bugs.
I wrote a message protesting this dreadful practice, but got a predictably barmy response from one of the devs. I'm not sure who he was speaking for, though.
Let's hope this doesn't catch on.
I used the word "deranged". He wasn't too happy about that.
@FaheemMitha sure yes some people do, some still use cvs, but all major code sites github/gitlab/bitbucket/sourceforge/... now steer people towards git and I don't see any real prospect of that changing any time soon, so i doubt many projects are using mercurial that were not set up a few years ago
@DavidCarlisle Yes, I know Git is now dominant, and that is unlikely to change in the foreseeable future.
09:48
@FaheemMitha you mean the bugs on mercurial itself not that they force that policy on your code if you use mercurial?
@DavidCarlisle Bugs relating to the mercurial project. I don't see how they could do this to anyone elses code.
I mean, the developers don't have god-like powers. Fortunately.
@FaheemMitha phew (just checking:-)
Still bad, though.
@FaheemMitha yes, i like my long list of 30 year old, still open bugs, they are old friends i wouldn't want a bot to kill them.
3
@DavidCarlisle Heh. Is this sarcasm? :-)
I'm not clear why old bugs don't get closed. Is it because they are too hard to fix? I.e. intractable. Or do people just not care enough to take the time to fix them? Or are other things just higher priority?
I mean, Mercurial is adding new features, so there are people willing to work on the project.
Overall, this attitude strikes me as distressingly similar to that of say, Microsoft.
09:52
@FaheemMitha well it's not exactly untrue:-) this seems to start in 1996 for example latex-project.org/cgi-bin/…
@DavidCarlisle I meant the "old friends" bit. I believe that there are old bugs.
And just to be clear, we're not talking about wishlist or unconfirmed bugs here. These are real bugs with reproduction recipes that get closed.
And people then run into them again, and report them again. It's all fun and games.
This actually happened to me last week. I wasn't too happy about it. Hence my email about "deranged practice".
Actually, this one latex-project.org/cgi-bin/ltxbugs2html?pr=tools/3479 has a suggested patch, but no response. Though the patch looks a bit unlikely to my inexpert eyes.
@FaheemMitha oh there are several like that, but they are unfixed for a reason. there are cases where the current spacing is clearly "wrong" and in some of the bug reports the pached code is probably "better" but if it's just a bit ugly and I change the spacing to be better but then 30 years worth of documents by millions of people start getting overprinting because they previously have \vspace{-something} to correct for the spacing inline, then it's not clear that fixing the bug is the right
... thing to do.
@DavidCarlisle Yes, I see your point. There's legacy code to think of.
I suppose one way to go is to create a brand new package. E.g. longtable2.
A better long table for the 21st century. With extra stretching power.
10:08
@marmot Quite
@DavidCarlisle I get this
tl2019pretest > luatex
This is LuaTeX, Version 1.10.0 (TeX Live 2019)
 restricted system commands enabled.
**^D
! End of file on the terminal... why?
@egreg yes I get that now after a cygwin update, I'm just wondering if I should flag it on tl list, but having updated I can't reproduce so It wouldn't be a very useful report, just that the new version has a dependency on a new version of one of the thousands of dlls that I just updated:-)
@FaheemMitha 5 not two but yes: github.com/davidcarlisle/dpctex/tree/master/longtable (one day...)
@DavidCarlisle 5?
@FaheemMitha the current release of longtable is version 4
@DavidCarlisle Ah. No, I mean change the name to longtable2.
So, a different package. Not the same package.
@FaheemMitha that is another possibility although simpler might be to wait for @JosephWright to write an l3table package so I don't need to do anything
10:16
@DavidCarlisle It's always good when others do the work for you.
What is l3table? Or does it currently not exist?
@FaheemMitha it doesn't exist but at some point something will be needed...
@DavidCarlisle I see. So for the moment longtable is the only game in town?
@FaheemMitha well supertable, tabu, xtab, ....
@DavidCarlisle Oh. I should check those out.
@FaheemMitha tabu is hmm a bit delicate at the moment github.com/tabu-fixed
10:20
@DavidCarlisle No immediate plans to switch away from longtable. It works for me.
@FaheemMitha good to know I have 1 happy user, at least.
@DavidCarlisle Is that version in TeX Live? I've never used tabu, though.
@DavidCarlisle But seriously, do you think closing old bugs, just because, is a good idea?
Or even an indifferent idea?
@FaheemMitha no, I think they should be open, even if you mark them as not to be fixed because "of whatever reason"
@FaheemMitha yes (well it's a version ahead but the version in texlive came from that repo)
@DavidCarlisle Ok. That's good.
 
1 hour later…
11:56
@DavidCarlisle Suppose you are on page 1 of a document. On page 2 you want to print a header, if a boolean value secondhead is set. Is it possible to set such a boolean on page 1 with for example \afterpage{\setboolean{secondhead}{true}} so that it remains stll false on page 1, but gets true on page 2?
@Kurt no, not really, but why not use marks or \thispagestyle both of which use built in tex features to get the right header on each page?
@DavidCarlisle I'm trying to understand the issue in this answer:
3
A: Repeating section headers in moderncv?

XavierStill one bug left, but hopefully someone else can figure it out :) Here is my idea: close the thebibliography environment at the start of every \bibitem to reopen it immediately. This way, LaTeX will print the bibliography head through \bibliographyhead between every bib item. Then it should b...

@Kurt blurg, the second para seems rather wrong, not motivated to read the code if it's implementing that:-)
The headline for bibliography with (Continued) should be repeated at the beginning of the page when starting a new bib entry (it can happen, that the last entry has to get finished first)
12:13
@Kurt so you should set a mark at the start and end of each entry, and then the page head can compare \topmark and \firstmark and if they are not the same the top of the page has an entry (given by \topmark) that started on the previous page.
@DavidCarlisle If you mean \setboolean...that can be changed, I simply do not know and could not find in the documentation of afterpage, if setting values of variables in which form ever is possible?
@Kurt no I mean "Here is my idea: close the thebibliography environment at the start of every \bibitem to reopen it immediately."
@DavidCarlisle Ah, I see. I will try. Thanks!
@Kurt \afterpage is just like \def or \newcommand it does not inspect the supplied tokens at all, absolutely anything can be placed there, but it is re-inserted and executed when tex starts making up the next page, and it may have already typeset some or all of that page, by that time.
@Kurt but using \afterpage to control page headings is somewhat weird as the page headings are the one place where you reliably know which page you are on, so controlling them from within the page by the highly invasive \afterpage macros is somewhat brave and optimistic.
@DavidCarlisle okay, then I understand why that try can not work as expected.
12:24
@DavidCarlisle ooh blurg
@DavidCarlisle I see that, yes. I guess the idea results from the thought, that the parameter of afterpage is executed firstly on the next page ...
@Kurt @Kurt remember of you have a very long paragraph it's all set in a line first (and so all macros and booleans and anything else are resolved at that point) then the line is broken in to paragraphs and only then page headings and after page tokens are inserted, the page heading comes in the head and the afterpage tokens are inserted by splitting the page and re-insertkng stuff inti the vertical list but in both cases a lot and perhaps all of the following page has already been typeset
@Kurt the places where \afterpage really executes are intersting such as if you have $$ then essentially it is executed between the dollars... the first $ has no effect` but when the second $ is seen tex exercises the page breaker so \afterpage may insert itself then, but this is before display math has started or \everydisplay has been inserted., it then has to force an internal page break so grabbing all of the page-so-far into a box then it
@Kurt restarts the main vertical list with its own tokens then the unboxed box that it grabbed from the page-so-far that actually triggered the page breaker, then anything that follows that. So the behaviour of any global settings you make in \afterpage` are well defined but not always what you might want, and typically not what you expect, unless you wrote the code:-)
@DavidCarlisle horrible details ;-)
@UlrikeFischer i'd never be brave enough to use afterpage in a real document
@UlrikeFischer meanwhile just the expected:
  Check failed with difference files
  - ../build/test/tlb-callbacks-001.luatex.diff
  - ../build/test/tlb-extraprimitives-001.luatex.diff
  - ../build/test/tlb-ltluatex-001.luatex.diff
@DavidCarlisle okay, I see. I thought it might be more complicated than it seems to be. Because I never heard of afterpage I just want to understand better that problem. Thanks for you explanaition!
@DavidCarlisle :-)
12:51
@Kurt try
\documentclass{article}

\setlength\textwidth{5cm}
\setlength\textheight{\dimexpr\topskip+10\baselineskip}
\usepackage{afterpage}
\sloppy

\begin{document}

\afterpage{HELLO\par\aftergroup\bfseries}

one two three four five six seven eight nine ten
one two three four five six seven eight nine ten
one two three four five six seven eight nine ten
one two three four five six seven eight nine ten
one two three four five six seven eight nine ten
one two three four five six seven eight nine ten
one two three four five six seven eight nine ten
the tokens are executed where it starts to go bold, but the typeset HELLO is re-inserted at the top of the page after half of page 2 has been typeset.
@DavidCarlisle Yes, I see. Good example!
 
1 hour later…
14:27
@barbarabeeton Ah, a mystery! That's okay, I don't need to know everything.
@HaraldHanche-Olsen ooh
14:59
Is there a way to output \msg_info:nn to the terminal, too, not only to the log (without it being a warning)?
15:11
@UlrikeFischer I see what you mean vvv
Failed tests for configuration build:

  Check failed with difference files
  - ./build/test/25-sec-non-ascii.luatex.pdf.diff
  - ./build/test/25-sec-non-ascii.pdftex.pdf.diff
  - ./build/test/38-pdfminorversion.luatex.diff
  - ./build/test/40-test-pdfhighlight.luatex.pdf.diff
  - ./build/test/40-test-pdfhighlight.pdftex.pdf.diff
  - ./build/test/o50-footnotesspaces.luatex.diff
  - ./build/test/test-bm-pu-licr.luatex.pdf.diff
  - ./build/test/test-bm-pu-licr.pdftex.pdf.diff
  - ./build/test/test0-pdf.luatex.pdf.diff
@DavidCarlisle ;-). For the pdftex failures due to the Producer entries, one should add something \AtBeginDocument{\let\@pdfproducer\relax} to regression-test, then hyperref will no longer try to add its entry (@JosephWright).
15:30
The first example of a document which fails in tl18 but works in pretest ;-)
@UlrikeFischer (@JosephWright) to be honest I don't know how to handle the pdf based tests, what are you supposed to do with diffs like ! <</D[38 0 R/XYZ 109.854 704.063 null]>> just look at the pdf and if it looks OK check in the new baseline tpf file?
@DavidCarlisle I see in the fc-files stuff like this:
***** .\BUILD\TEST\25-sec-non-ascii.luatex.tpf
    9:  5 0 obj
   10:  <</S/GoTo/D[6 0 R /Fit]>>
   11:  endobj
***** .\BUILD\TEST\25-SEC-NON-ASCII.LUATEX.PDF
    9:  5 0 obj
   10:  << /S /GoTo /D [ 6 0 R /Fit ] >>
   11:  endobj
*****
And it is rather obvious that the difference is only in white space.
15:45
@UlrikeFischer hmm I suppose l3build could normalize that away if it is also removing the byte offsets, but still seem to be rather a lot of diffs
@DavidCarlisle I don't think that it would worth the time to do something in l3build - they probably won't change the spacing with the next luatex version again. I wonder if there is an option which removes the spaces again. But I can look at this diffs and resave the tests when we switch to tl19.
I just got a palindrome.
@JouleV I suppose I am duty bound to point out it would look more palindromic if you added 440000 more points.
16:02
@DavidCarlisle I have to edit the code a bit then
user image
4
@JouleV excellent palindrome!
@DavidCarlisle Luckily the score only has 4 digits. If it has 5 digits I can do nothing ;-)
@JouleV aim for a central digit of 0 or 8 :-)
@DavidCarlisle Oh I just found a solution. I will delete the comma instead of replacing it.
@PauloCereda -- Neat! Thanks!
16:12
Bonjour Everyone
I want to typeset Punjabi but don't know how?
@JouleV although once you start down that path there is temptation for corruption and alteration of the facts. Not that I would every do anything like that. By coincidence i just got a palindrome too:
user image
4
@PrabhjotSingh most likely you will find it easiest to use xelatex, I will see if i can find an example...
@DavidCarlisle LOL :))) Jon Skeet is beaten :D
Yes this is only a tricky way. If you want to have a perfect palindrome your score must have 1, 2, 3, 6, or 12 digits and so on...
@PrabhjotSingh something like this (if google translate made something readable)
\documentclass{article}
\usepackage{fontspec}
\setmainfont{Nirmala UI}
\begin{document}

ਇੱਕ ਦੋ ਤਿੰਨ. ਸਤਿ ਸ੍ਰੀ ਅਕਾਲ ਦੁਨਿਆ

\end{document}
or a mixed english/punjabi example here:
4
A: How can I use a Gurumukhi font to typeset Punjabi with XeLaTeX or PDFLatex?

Alan MunnHere is a possible solution using the Saab font. I don't know anything about the Gurmukhi script, and the Saab font doesn't seem to recognize the Gurmukhi script tag, even though it should, so there may be problems with the font. For small sections of Punjabi text, this may be sufficient. For wh...

16:37
@DavidCarlisle Fatal fontspec error: "cannot-use-pdftex". Fontspec package requires either Xetex or LuaTeX.
@PrabhjotSingh as it says: use xelatex or lualatex.
@PrabhjotSingh yes you need to use xelatex not pdflatex
@PrabhjotSingh with xelatex (or lualatex, but xelatex probably better here) you can use any of your normal system fonts, so if you have a font that you use in a browser or word processor you can use it in xelatex. pdftex only uses the classic tex font system so any font has to be prepared specially for tex and can have at most 256 characters, and pdftex can not do any of the complicated font shaping required by at least some languages of the region.
@DavidCarlisle Xelatex gives ! Undefined control sequence. And .pdf file reads article fontspec UI document.
@PrabhjotSingh show the actual error from the log, so we can see which command was undefined. (if it is \documentclass make sure you used xelatex not xetex)
16:56
This is XeTeX, Version 3.14159265-2.6-0.99998 (TeX Live 2017/Debian) (preloaded format=xetex 2019.3.2) 2 MAR 2019 22:09
entering extended mode
restricted \write18 enabled.
%&-line parsing enabled.
**test.tex
(./test.tex
! Undefined control sequence.
l.1 \documentclass
{article}
?
! Undefined control sequence.
l.2 \usepackage
{fontspec}
?
! Undefined control sequence.
l.3 \setmainfont
{Nirmala UI}
?
! Undefined control sequence.
l.4 \begin
{document}
?
Missing character: There is no ਇ in font cmr10!
@PrabhjotSingh as I guessed you are using format=xetex so plain tex not latex. use xelatex not xetex, you would get the same error if you used pdftex on a pdflatex document.
@DavidCarlisle This is from pdflatex.
his is pdfTeX, Version 3.14159265-2.6-1.40.18 (TeX Live 2017/Debian) (preloaded format=pdflatex 2019.3.2) 2 MAR 2019 22:28
entering extended mode
restricted \write18 enabled.
%&-line parsing enabled.
**test.tex
(./test.tex
LaTeX2e <2017-04-15>
Babel <3.18> and hyphenation patterns for 84 language(s) loaded.
(/usr/share/texlive/texmf-dist/tex/latex/base/article.cls
Document Class: article 2014/09/29 v1.4h Standard LaTeX document class
(/usr/share/texlive/texmf-dist/tex/latex/base/size10.clo
File: size10.clo 2014/09/29 v1.4h Standard LaTeX file (size option)
@PrabhjotSingh when you do get that going you may find that you don't have Nirmala UI font as it's probbaly a windows one.
@PrabhjotSingh yes you can not use these fonts with pdftex as we said before, what I mean is that if you take a pdflatex document and try to use pdftex you will get the \documentclass undefined error.
@DavidCarlisle Agreed, I been doind that for two hours.
@PrabhjotSingh but you can replace Nirmala UI by any font that you would use with firefox or open office etc (I picked that one as that was the font firefox used on windows for the text I had translated)
@PrabhjotSingh so use xelatex and all should be fine:-)
@PrabhjotSingh google suggests you should have the lohit font in Debian so save the following as test.tex and then type the command xelatex test in the terminal in that directory and you should get a pdf file showing the text (which i hope says something, I can not read the script)
\documentclass{article}
\usepackage{fontspec}
\setmainfont{Lohit-Punjabi.ttf}
\begin{document}

ਇੱਕ ਦੋ ਤਿੰਨ. ਸਤਿ ਸ੍ਰੀ ਅਕਾਲ ਦੁਨਿਆ

\end{document}
17:21
@DavidCarlisle the english translation looks ok: one two Three. Hello World, but german is rather odd: Eins oder zwei drei Sa Sri Akal Dionya ;-)
This is XeTeX, Version 3.14159265-2.6-0.99998 (TeX Live 2017/Debian) (preloaded format=xelatex 2019.3.2) 2 MAR 2019 22:49
entering extended mode
restricted \write18 enabled.
%&-line parsing enabled.
**test.tex
(./test.tex
LaTeX2e <2017-04-15>
Babel <3.18> and hyphenation patterns for 84 language(s) loaded.
(/usr/share/texlive/texmf-dist/tex/latex/base/article.cls
Document Class: article 2014/09/29 v1.4h Standard LaTeX document class
(/usr/share/texlive/texmf-dist/tex/latex/base/size10.clo
File: size10.clo 2014/09/29 v1.4h Standard LaTeX file (size option)
@UlrikeFischer Google Translate would be useful after say 50 or 70 years later,imo.
@PrabhjotSingh Ok so do you know a font that you do have?
@PrabhjotSingh well it was useful for me just now to generate a test phrase (how close to being correct was it:-)
@UlrikeFischer oops if I go to any page on the main site, the firfox tab crashes:-)
17:41
@JouleV Do you have a tikz code for making the user profiles more catchy? One could use gold badges/.initially=5 and so on to make them accessible to everyone. ;-)
18:01
@marmot Hi! Do you mean the TikZ code for the reputation board?
@JouleV I meant a TikZ code that allows one to move the comma in a score or to "improve" the score.
@marmot I don't think this is really related to TikZ though :-)
@JouleV Of course not but then we could see how far @DavidCarlisle will get with the picture mode. ;-)
@marmot I'd probably use javascript.....
@DavidCarlisle That's considered cheating.
18:09
@marmot I have very catchy user profiles:
user image
4
@UlrikeFischer One word: AMAZING!!!
@UlrikeFischer Perfect! Well, almost. I think the silver badges should become koalas and the gold badges marmots. ;-)
Each site can have its own set of badge symbols. This is computer science.se's
Why doesn't TeX.SE have tikzlings and tikzducks as the set?
@JouleV Yes. Bronze: duck, silver: koala, gold: marmot. ;-)
@marmot we would need @SamCarter for this change, she made the picture I'm using to overload the badges. And the gold badges must be ducks.
18:15
@UlrikeFischer But then everyone would just try to go for the bronze badges only. ;-)
@marmot but using the dom inspector and just editing the content inline is not, I assume.
@DavidCarlisle Why?
@marmot see it this way: if marmots are bronze there will much more marmots than ducks around.
@DavidCarlisle I see. That's a good way of avoiding violating the laws: just change them.
@UlrikeFischer But this is completely unrealistic.
18:19
@marmot Using the inspector one can even become the POTUS.
@UlrikeFischer it could look more like a menu, gold, duck with orange sauce, silver, duck pie, bronze, crisply duck pancakes.
@JouleV Yes, and change the laws that one may violate otherwise after one is the POTUS...
@DavidCarlisle How can I define such a \topmark and \botommark for bibitem? I found package titlesec having defined a \topmark, but I do not know if I can use that. Do you have a tip for me?
@Kurt basically you use \markboth{...} with the required heading for the item at the start and end of each item, and then access them in the header
@DavidCarlisle okay, I will try to use it. Lets see ...
18:46
@DavidCarlisle I know why I didn't show how your user profile looks here.
user image
4
@marmot ^^^ If you want them add this to your css:
.badge1, .badge2, .badge3 {
         background-image: url("https://raw.githubusercontent.com/samcarter/shared/master/TeXse/sprites2.svg?sanitize=true") !important;
      }
3
19:21
@UlrikeFischer ;-)
@marmot thank @samcarter ;-). She was fast.
@UlrikeFischer Yes, that's what I thought. @samcarter Mille grazie!
@marmot And while ducks looks in good in every color, I don't think that a marmot can wear gold ;-)
@UlrikeFischer But marmots smell better... ;-)
 
1 hour later…
20:43
Jean Grey kills Mistique in Dark Phoenix!
20:58
@JosephWright do you know if there is something equivalent to \pdfvariable majorversion and \special{pdf:majorversion 2} in pdftex ? (to get pdf 2.0 and higher)?
@UlrikeFischer Current no, we'd have to request one (i.e. I'm looking at the sources now!)
@UlrikeFischer it writes a literal string %PDF-1. before looking at any variables...
@JosephWright source isn't very flexible:-) pdf_print("%PDF-1.");
@DavidCarlisle Yeah, but the next line is pdf_print_int_ln(fixed_pdf_minor_version);, which is presumably what we need for the major version too
@DavidCarlisle Like I said, we presumably need to ask for another new primitive (@UlrikeFischer)
@JosephWright yes replace the 1. by a variable but also the logic just above that needs to be guarded so if (fixed_pdf_minor_version >= 5) becomes if (fixed_pdf_minor_version >= 5 or fixed_pdf_major_version > 1 )
@JosephWright I'm assuming there wouldn't be time to get this asked for and tested in time for tl2019
@DavidCarlisle Yes, sure. But my point is we already have 95% of the code we need in the minor version stuff. It's pretty mechanical to extend to allow major_version as 1 or 2
@DavidCarlisle Ask KB?
@DavidCarlisle It only took a day for the pTeX bug I found to be fixed ...
@DavidCarlisle We have three weeks, and they do binary rebuilds ...
21:05
@JosephWright with latex + dvipdfmx it works via \special{pdf:majorversion 2} or \special{dvipdfmx:config V 2.0}. What about dvips? How it the version set there?
@JosephWright I suppose so, this came originally from a comment made in hyperref but there it's similar, easy to add an option to set to 2 but harder to find all the places where it's testing for minorversion<5 and removing features
@UlrikeFischer Er, that's probably tricky. A lot of this comes down to command line switches for ps2pdf (for example compress level is done that way)
@JosephWright I think we should try.
@DavidCarlisle Yeah
@UlrikeFischer, @DavidCarlisle I'll mail Karl, we should talk on Monday
@JosephWright Ok thanks
I've got no firefox:(
21:07
@JosephWright No I meant generally. How is e.g. pdfversion 1.4 or 1.5 set when using the dvips/ps2pdf route? I think I never did it ;-)
@UlrikeFischer set by ghostscript I think (ps2pdf_
@UlrikeFischer the ps2pdf script manual says:
The three scripts differ as follows:

- ps2pdf12 will always produce PDF 1.2 output (Acrobat 3-and-later
compatible).

- ps2pdf13 will always produce PDF 1.3 output (Acrobat 4-and-later
compatible).

- ps2pdf14 will always produce PDF 1.4 output (Acrobat 5-and-later
compatible).

- ps2pdf per se currently produces PDF 1.4 output. However, this
may change in the future.
@UlrikeFischer so back to the future part II...
@DavidCarlisle, @UlrikeFischer Mail sent (see team list)
@DavidCarlisle Right, so do you want to take this one?
@JosephWright oh the web, i used man ps2pdf :-)
21:10
@DavidCarlisle No man here
@JosephWright same text anyway
@DavidCarlisle No PDF in C:\texlive\2018\texmf-dist\doc\man\man1 for ps2pdf :(
@JosephWright man here just uses the classic nroff terminal output, no pdfs
@DavidCarlisle Have the .1 files for all PDFs (texdoc uses PDFs, man uses the .1 files if you have `man): I guess I could set it up ...
@UlrikeFischer @JosephWright :
$ file sample2e.pdf
sample2e.pdf: PDF document, version 2.0
21:16
@DavidCarlisle Done already?
@JosephWright buh blaming me ;-) Not quite sure if it is really necessary for tagging (that's hard in pdftex anyway) even if some stuff exists only in the 2.0 reference, but it would make hyperref easier, if one doesn't need different version code for pdftex.
that is with latex sample2e; dvips sample2e; ps2pdf -dCompatibilityLevel=2.0 sample2e.ps
@DavidCarlisle Ah
@JosephWright I compared the shell scripts for ps2pdf12, ps2pd14 and extrapolated:-)
@DavidCarlisle when I do this ghostscript says
GPL Ghostscript 9.26:
It is not possible to preserve transfer functions in PDF 2.0
transfer functions will be applied instead
Drücken Sie eine beliebige Taste . . .
I wonder what this means ;-)
21:19
@UlrikeFischer yes it says same for 1.9 I assumed it was true
@UlrikeFischer Out of cheese error
@DavidCarlisle @JosephWright but if all other routes can do 2.0 pdftex should really get it too ...
@UlrikeFischer gs has a page that says: PDF 2.0 deprecates the use of transfer functions, and so when producing PDF 2.0 compatible output if the TransferFunctionInfor is set to /Preserve it will be silently replaced with /Apply. You can instead specifically set TransferFunctionInfo to /Remove when producing PDF 2.0 in order to avoid the transfer function being applied.
@DavidCarlisle All of the validation is in that one procedure, so I think adding major version support just needs us to add variables/primitives, adjust there plus I guess adjust the .ini to explicitly set v1
@DavidCarlisle Should we resist to adding some witty comment to tex.stackexchange.com/q/474765/4427?
21:26
@JosephWright yes I think so (and same in uptex....)
@DavidCarlisle upTeX doesn't have PDF output
@egreg I saw the question at the time and decided not answer:-)
@DavidCarlisle these are too long sentences for this late hour and the wine level. ;-) Do I need to check what transfer functions are?
@JosephWright ah of course, and xdvipdfmx
@UlrikeFischer probably not, they are deprecated pdf level 1 thing (I think I may have known once but not at this minute) so being a new modern pdf 2 devotee you can forget they existed.
@DavidCarlisle @UlrikeFischer said \special{pdf:majorversion 2} ...
21:29
@JosephWright or \special{dvipdfmx:config V 2.0}
@UlrikeFischer Yeah, that too
@JosephWright so she did, so just pdftex then...
@DavidCarlisle Yup
@UlrikeFischer I guess I'd use that: it's clear it is about dvipdfmx
@DavidCarlisle Like I said, do you want to take this, or should I?
@JosephWright probably safer if you did, I half watched the expanded changes but I'm sure you would be quicker and more accurate with the web changes.
@DavidCarlisle Oh great (!) Now I'm the WEB expert on the team?
21:36
@JosephWright assuming Karl doesn't come back and say no chance
@DavidCarlisle I'll have a .diff shortly
@JosephWright thanks
@DavidCarlisle Updating my fork on GitHub: I want to be sure things build!
21:54
@JosephWright btw I wrote today also to the luatex list: setting there the version to 2.0 and then using objcompresslevel=1 gives the warning objcompresslevel > 0 requires minorversion > 4.
@UlrikeFischer Ho ho
@UlrikeFischer I suspect there are tests all over the place in multiple pieces of software that are only checking the minor version
@DavidCarlisle Yup
@DavidCarlisle Quite probably. And I don't assume that there is a garanty that "2.0" really gives 2.0, but the standard is there since 2017 so some steps towards it would be good. And if the header is there, I can check with preflight.
@DavidCarlisle, @UlrikeFischer Nearly done but ... there is a lot of stuff about reading images that I'm not sure how to handle
22:13
@JosephWright yes lots of tests (eg in writepng.c) with fixedpdfminorversion <= 1 it's tempting to make the internal integer fixedpdfminorversion = 10 for pdf 2.0 then all those integer tests do the right thing
@DavidCarlisle Not that exactly: see pdftex.web @<Undump pdftex data@>= and so on
@DavidCarlisle That is a pretty big issue :(
@JosephWright tl2020?
@DavidCarlisle Probably ... we can't really half-arse patch this
@DavidCarlisle We might want to try to organise a 'general review' of all PDF generation code in TL in this sense: there are probably lots of issues
@JosephWright yes
@DavidCarlisle I'll forewarn KB
22:23
@JosephWright hm. I wonder if stuff like this is hidden in luatex too.
@UlrikeFischer Yes, I suspect so. Drafting mail to KB now
@DavidCarlisle Same for LuaTeX
@JosephWright can't say I'm surprised:-)
@UlrikeFischer It is: the C files look much like the pdfTeX ones
@DavidCarlisle See my mail to KB
@JosephWright if pdf_major_version = 2 then begin future proof as if pdf_major_version > 1 then begin ?
@DavidCarlisle Like I think I've said before, issues in the C code worry me a lot less than in the WEB: lots of people read/write C
22:34
@JosephWright I'm starting to feeling guilty ;-) I opened a can of worms by looking at the hyperref driver files.
@UlrikeFischer No, this stuff is important
@JosephWright something that should be sent to the luatex list?
@UlrikeFischer I guess; might be worth first trying to get a feel on the scale of the issue

« first day (3044 days earlier)      last day (2109 days later) »