@AlexG I think my mail regarding setting the pdfversion for ghostscript didn't make it to gs-devel. Could you ask if it could be set with some postscript code?

@UlrikeFischer Sure, I could. Would you please provide me with a few arguments that I can use to emphasise the need for this?

@AlexG mail sent.

@DavidCarlisle Do we have any docs for \DeclareRelease, etc.?

@UlrikeFischer Great, thanks!

@UlrikeFischer Is it ok to tell them that I am writing on your behalf?

@AlexG sure.

@UlrikeFischer Done.

@AlexG thanks. I hope I don't forget to keep track of this ;-)

@UlrikeFischer I am curious too about what they will answer. Hope we will get a reply soon. (Some of the developers are Britain-borne, as far as I know.)

@JosephWright harder to find than it ought to be (see email discussion I had with Frank on team list while adding these to longtable) but ltclass.dtx or latex-project.org/publications/…

@DavidCarlisle Yes - I've ended up copying stuff from array

@DavidCarlisle quack

@PauloCereda \DeclareCurrentBreakfast

@DavidCarlisle oh no

@UlrikeFischer ooh we ducks are very good at keeping tack of things

@UlrikeFischer answer:

--- error --- Improper op while scanning font defs in postamble
[FATAL]   make4ht-lib: Fatal error. Command tex4ht returned exit code 1

@DavidCarlisle it's not my fault

@PauloCereda but I have proof:

@DavidCarlisle wait, what?!

@UlrikeFischer You have got mail.

@PauloCereda <3

@DavidCarlisle github.com/josephwright/siunitx/commit/…

@AlexG <3

@JosephWright oh no, more time travel

@DavidCarlisle I notice the 'name' part is quite awkward to use - as we don't enforce any particular type of version strings, I can't (readily) specify 'version 2' there, only 'v2.8d'

@JosephWright Stop harassing @PhelypeOleinik!

@PauloCereda Oh, I'm hassling FMi

@JosephWright oh in that case, carry on :)

@JosephWright could you release siunitz?

@AlexG I thought we tried <</CompatibilityLevel 2>> setpagedevice before? (@UlrikeFischer)

@UlrikeFischer Oooh

@UlrikeFischer So I should add that to the backend?

@JosephWright well we should at first try if it works. I thought we did and it didn't work, but perhaps we were to late?

@JosephWright the less people know, the better :)

@DavidCarlisle how did you manage to read my mail?

@UlrikeFischer ooh

@JosephWright you can use v2 it doesn't have to match the actual decalted versio at the time it was release, but you can't do any comparisons on that other than equality so if the user uses [=v2] they get exactly that not that-or-later as you can do with teh date version

@DavidCarlisle Yes, methought the same.

@UlrikeFischer I'm big brother's little brother

@DavidCarlisle but still old experienced? :)

@DavidCarlisle Just spotted that - I've added "v2" (doesn't help that the code doesn't treat v2 and 2 the same, as the 'v' is not formally part of the version number)

or maybe I read this: ghostscript.com/pipermail/gs-devel/2021-April/010422.html

@DavidCarlisle I'd just rather have \DeclareRelease{2.8d}{2021-04-09}{siunitx-v2.sty} know that =v2 or=2.8 should still match

@JosephWright is it worth keeping a compatibility layer?

@PauloCereda Here, yes - when I made the v1 to v2 change, I initially didn't do that - it was a bad idea

@JosephWright I (think) you could have three DeclareRelease with all three strings but as the "name" part follows no pattern and just has the names you document I don't see it helps you could go for a traffic light system and call it red amber green then no one would worry about dates at all and be happy.

@JosephWright oh

@PauloCereda I know I will break some things here: I'm switching to semantic versioning so the whole point is it's a breaking change

@DavidCarlisle I guess it's simply not worth it: unlike some other languages, we don't enforce any version string system

@JosephWright which is good. :) I'd rather have a migration section in the manual (or even in a separate document) for quick fixes and never speak or maintain v2 again. :)

@PauloCereda I have that - I'll dig out the GitHub link

@JosephWright wanna hide in the island? :)

@PauloCereda github.com/josephwright/siunitx/blob/…

@JosephWright ooh you are efficient

@PauloCereda Does it look OK?

@UlrikeFischer So more generally we want <</CompatibilityLevel [major].[minor]>> setpagedevice?

@UlrikeFischer @JosephWright can we blame Will?

@DavidCarlisle Yes, sounds good

@JosephWright I suppose I have to try to find out why it's touching combining accents..

@JosephWright hmm I wonder what this does...

\cs_set:Nn \__um_set_mathsymbol:nNNn
 {
  \bool_lazy_and:nnT
   {
    \int_compare_p:nNn {#4} > {127}
   }
   {
    \int_compare_p:nNn { \char_value_catcode:n {#4} } = {11}
   }
   { \char_set_catcode_other:n {#4} }

@JosephWright I have a feeling we may have tried 2.0 before, perhaps it has to be 2 currently

@DavidCarlisle Ah, right

@DavidCarlisle I was wondering about the 1.x series - does that work?

@DavidCarlisle I'll wait for @UlrikeFischer to test :)

@JosephWright I can't remember what we were trying but what we tried worked for 1.x for x between 1 and 7 :-)

@DavidCarlisle Ah, so perhaps we need [major][if minor <>0 .minor]

@JosephWright version strings are tricky

@DavidCarlisle Yes

@JosephWright /shivers

@UlrikeFischer @DavidCarlisle : I ran a test with

\special{! <</CompatibilityLevel 1.4>> setpagedevice}
hello
\bye

It only works if we replace %!PS-Adobe-2.0 with %! in the PS output from dvips.

Should I ask on gs-devel?

@AlexG Ah

@AlexG Please

@JosephWright it does!

@PauloCereda Cool

@PauloCereda I'll be prodding everyone for beta testing soon enouhg

@AlexG wow, that really seems to work, even with ps2pdf.

@JosephWright Cool

@AlexG We could manage that, I guess ...

@JosephWright Nevertheless, I am going to ask whether the different behaviours are on purpose.

@AlexG Yes, does seem reasonable - see the recent business with TeX Live on Windows and our auto-detection of platform, for example

\special{! <</CompatibilityLevel 2.0>> setpagedevice} works to then, so we can set the major.minor always, and it wins over command line options.

@UlrikeFischer Oooh

@UlrikeFischer So that would be easy enough to add?

@UlrikeFischer but still editing the %!PS comment" ?

@DavidCarlisle yes, you have to remove this first. I must say I would never had guessed that this could lead to problems.

@UlrikeFischer does %%!PS (but not EPSF; comments have been disabled) work? (dvips has an option to make that already)

   findpapersize();
   if (disablecomments) {
      fprintf(bitfile,
             "%%!PS (but not EPSF; comments have been disabled)\n");
      fprintf(stderr, "Warning:  no %%%%Page comments generated.\n");
   } else {
      if (multiplesects) {
         fprintf(bitfile,
             "%%!PS (but not EPSF because of memory limits)\n");
         fprintf(stderr, "Warning:  no %%%%Page comments generated.\n");
      } else {
         isepsf = epsftest(sect->bos);
         if (isepsf)
            fprintf(bitfile, "%%!PS-Adobe-2.0 EPSF-2.0\n");

@DavidCarlisle yes that works : %%!PS-Adobe-2.0

%!PS-2.0 and %!PS-3.0 work too. Just "-Adobe" is too much.

Acc. to adobe.com/content/dam/acom/en/devnet/actionscript/articles/…, pg. 17, conforming documents start with %!PS-Adobe-3.0 and those starting with %! are explicitly said to be non-conforming, on page 21.

@DavidCarlisle but I think the disablecomments and multiplesects option would have more side effects, so probably not the best solution.

@AlexG so you think this is a gs feature they may fix or should we get dvips to have an option not to write Adoibe ?

@DavidCarlisle %!PS-Adoibe-3.0 would work too

@UlrikeFischer :-D

@DavidCarlisle Doing this for dvips could be some kind of resort, if Artifex won't be willing to change this.

@DavidCarlisle Let's wait for the reply on gs-devel.

@LukeTheWolf sure but it's best to make a small complete example showing your issue and asking a question on the main site (not here in chat) discussing code fragments in chat is not as easy (despite me posting some code just above)

Two disposable gloves tied, full of hot water, simulating impossible human contact in a Covid isolation ward nearby my city.

@PauloCereda Oh my

I could provide you some code, but I think that it would be useless in this case

@DavidCarlisle regarding XeTeX fonts, see kodymirus.cz/tex4ht-doc/Configurations.html#x8-660007.14.2

@LukeTheWolf I think I would probably just try to create the L and M boxes as rectangular nodes with a set size, and then place the circle nodes in the appropriate places without content.

@michal.h21 aha that looks promising!

@UlrikeFischer ^^^ do you think this reading documentation idea might catch on?

it is necessary to instruct TeX4ht to load non-latin Unicode codeblocks. This can be done either using Script option of \newfontfamily, or using \xeuniuseblock in the .cfg file. Like this:

\Preamble{xhtml}
\ifdefined\xeuniuseblock
\xeuniuseblock{Greek}
\xeuniuseblock{CJK}
\fi
\begin{document}
\EndPreamble

@LukeTheWolf There's the TikZ automata + fit libraries that can easily help you achieve something similar. I've done lots of things like this. :)

@michal.h21 for the learnlatex interface I'd really like it not to have to rely on specifics in the document other than the package configurations, so how bad would it be if I always did that if fontspec was loaded with xetex ?

@DavidCarlisle yes, I think it is easiest to use LuaTeX instead of XeTeX in most of cases. Unicode is supported by Lua callbacks, so you don't need to deal with Unicode block loading. You will also save some memory by not making thousands of characters active :)

@michal.h21 that looks like a very nice docu. Did you make it?

@PauloCereda I'm actually using TikZ automata, what if I post here my snippet? Maybe it could be interesting for giving me more specific hints !

@DavidCarlisle I think it is OK. you can add the \ifdefined\xeuniuseblock part to the universal config file. it should be used only with xetex.

@michal.h21 OK thanks

@LukeTheWolf you can try. :) But perhaps it might be better suited for the main site.

@UlrikeFischer yes. it is generated automatically by github actions every time I push changes to github.com/michal-h21/tex4ht-doc

@PauloCereda Ok! Thanks

@LukeTheWolf My pleasure! Sadly, at the end of the day, tweaking is always required, so there's no magic happening. Let me see if I remember another package...

@DavidCarlisle that would be great. the current tex4ht command, that converts DVI, has a bug, that we don't know how to fix, and it fails with a fatal error every time it cannot find a tfm file for a font. there is an alternative version that supports open type fonts, but it is not stable yet: github.com/mingiss/tex4ht-vtex

@michal.h21 "making active" means you redefine the chars to write some special representations for the chars to avoid to have chars >255 in the dvi, as the binary used in the following step can't handle them?

@UlrikeFischer yes, exactly. this is why only Latin script is handled by default.

bbc.com/news/uk-11437314

@samcarter_looks_forward_TUG'21 oh my

@PauloCereda he was the most funny one of them :(

@samcarter_looks_forward_TUG'21 oh shame he didn't quite make his 100

@DavidCarlisle so close!

@LukeTheWolf Suggestion: add \usetikzlibrary{fit} to your preamble and try something like \node[draw, fit={(1)(2)}] {}; somewhere in your TikZ picture. :)

The rest is tweaking stuff. :)

@DavidCarlisle , @Ulrike , @JosephWright Acc. to Ken's reply, if I correctly understood it, in a DSC-conforming document, <</CompatibilityLevel .\..>> setpagedevice would have to be put before the first valid DSC comment to have effect, that is, even before %!PS-Adobe-2.0, because %!PS-Adobe-2.0 starts a conforming document which causes gs to immediately open the output file.

@AlexG yes sounds like that. Which probably means that one need to change dvips? Unless dvips has something that would allow to write before the comment.

@DavidCarlisle does that happen with some example from these? github.com/davidcarlisle/latexcgi/tree/main/make4ht

@PauloCereda It seems to work properly! Thx a lot!

@LukeTheWolf Will draw something for you, in a minute

@PauloCereda Great! I really appreciate it!

@PauloCereda thanks for mentioning this library: you just solved a problem I did not know I had :)

@samcarter_looks_forward_TUG'21 ooh <3

Hello everybody, a quick question: How can I move the label text, so that they are closer to the nodes (circles)?

I'm trying to install TL2021 on Windows10 but I keep getting an error about an undefined subroutine @TeXLive::TLWinGoo::is_vista. That's the first time I've ever had issues with the TL installation, does somebody have experience with this?

@LukeTheWolf vvv

\documentclass{article}

\usepackage[T1]{fontenc}

\usepackage{tikz}
\usetikzlibrary{automata}
\usetikzlibrary{positioning}
\usetikzlibrary{fit}

\begin{document}

\begin{tikzpicture}[
  shorten >=1pt,
  node distance=2cm,
  on grid,
  auto
]

\node[state, initial] (q0) {};
\node[state] (q1) [right=of q0] {};
\node[state] (q2) [right=4 of q1] {};
\node[state, accepting] (q3) [right=of q2] {};

\node[
  draw,
  dotted,
  rounded corners=2pt,
  inner sep=6pt,
  fit={(q1)(q2)}
] (L) {};

\node[right=of q1] {{\sffamily\bfseries\large L}};

@PauloCereda Holy moly! This is perfect! Thank you so much

@LukeTheWolf My pleasure. :) Take a look at the TikZ manual (although it's a bit daunting, I know) for some interesting tweaks, like inner seps, spacing, positioning, it helps a lot. :)

@PauloCereda I don't wanna bother you but have you got any idea of how can I do the curved line in this picture? ibb.co/wgkx7JM

@LukeTheWolf I cannot remember by heart now, but there's an in and out thingies that you can tweak...

@PauloCereda Of course, I will (when I finish the exams palmface)

@LukeTheWolf Good luck with your exams!

@PauloCereda No worries!

@PauloCereda Thx!

@LukeTheWolf ^^ something like this would help

@GermanShepherd IIRC there's a pos key that could be used...

@UlrikeFischer But I doubt, that PS with <<...>> setpagedevice before %!PS-Adobe-3.0 is still a conforming document...

@AlexG it probably then acts like the %! version won't it?, I think the %!PS-Adobe.. comment has to be the initial bytes in the file to have any effect?

@DavidCarlisle Yes, that's what the Adobe document on DSC says. But if setpagedevice is emmited as the second output line it is too late for GS. (And even then, it will break conformity, as the header section must consist of DSC comments only.)

After all, it seems that setting the PDF C level through PostScript is impossible without breaking conformity.

@AlexG seems that way. seems a bit rubbish. If latex/dvips would call dvips automatically the way xetex/xdvipdfmx does, then latex could control the dvips command line....

@DavidCarlisle Infact, the ps2pdf command line would need to be under tex's/dvips' control as this is obviously the only location where to set the PDF level.

@AlexG or perhaps simpler texlive could supply an alternative script to ps2pdf that greped the compatiblitylevel ... setpagedevice and passed the equivalent option to ps2pdfwr

@AlexG oh sorry yes half my brain had that idea while other half thinking of dvips commandline:(

@DavidCarlisle Yes, this could be an option.

@DavidCarlisle if we do it like this any comment would do, it doesn't need to be setpagedevice which is ignored anyway.

@UlrikeFischer yes

@UlrikeFischer if you have grep and sed this works I think

#!/bin/sh
# Convert PostScript to PDF.

# Currently, we produce PDF 1.4 by default, but this is not guaranteed
# not to change in the future.

pdeviceversion=`grep '<<.*CompatibilityLevel .*setpagedevice' $1 | sed -e 's/.*Level *\([0-9.]*\).*/\1/'`

echo $pdeviceversion

if [ "X$pdeviceversion" = "X" ] ; then
    version=1.4
else
    version=$pdeviceversion
fi


exec  ps2pdfwr -dCompatibilityLevel=$version "$@"

Watching the BBC

@PauloCereda Limited options

@PauloCereda I watched the 1 o'clock news but back here now

@JosephWright oh

@DavidCarlisle I am watching the 11AM news :)

@michal.h21 not with the example as written as you disable fontspec font interface to stop it so I was experimenting with what happened if you loaded the font anyway so I used the primitive \font directly (and found out why you disable things:-) vvv

% !TeX make4ht
% !TEX xelatex
\documentclass{article}
\usepackage{fontspec}
\setmainfont{texgyretermes-regular.otf}
\newfontfamily\cjkfont{FandolSong-Regular.otf}
\begin{document}

\font\rm=[texgyretermes-regular.otf]  \rm
\font\cjkfont=[FandolSong-Regular.otf]

ABC → αβγ → {\cjkfont 你好}

\end{document}

@DavidCarlisle % arara: learnlatex: { backbone: texlive }

@UlrikeFischer Did you check how Distiller handles PS that tries to set the PDF version?

@PauloCereda you got a mild rebuke:

I spotted an invalid directive (1) in the provided file. Make
sure to fix the directive and try again.

@DavidCarlisle pesky tool, I wondered who wrote it

@DavidCarlisle (you probably wrote it wrong, the error should be a different one :)

I could not find a rule named "learnlatex" in the provided rule
paths. Perhaps a misspelled word? I was looking for a file named
"learnlatex.yaml" in the following paths in order of priority:

@AlexG no, but I can try later. We are currently out.

@UlrikeFischer Oh, have a nice walk :)

@DavidCarlisle ooh :)

@DavidCarlisle pesky tool, I wondered who wrote it

@PauloCereda do you know what happens if I try % arara: xelatex ?

@DavidCarlisle It will run xelatex? :)

@PauloCereda texchnically it runs xelatex on something, yes

@DavidCarlisle ooh

$ arara test-xetex-fonts.tex
  __ _ _ __ __ _ _ __ __ _
 / _` | '__/ _` | '__/ _` |
| (_| | | | (_| | | | (_| |
 \__,_|_|  \__,_|_|  \__,_|

Processing "test-xetex-fonts.tex" (size: 335 B, last modified:
04/09/2021 15:24:37), please wait.

(XeLaTeX) XeLaTeX engine ................................   ERROR

I could not run the provided system command, something bad
happened. This part is quite tricky, since it involves aspects of
the underlying expression language. I will do my best to help you
in any way I can. There are more details available on this

@PauloCereda ^^ basically you get that on any file.

@DavidCarlisle Cygwin woes. :)

% arara: xetex: { options: [ '--fmt=xelatex' ] }

@DavidCarlisle ^^ try this one

    (XeTeX) XeTeX engine .................................... SUCCESS

@DavidCarlisle cool

@PauloCereda I never understood why cygwin never got a native java, mapping between windows java and a cygwin shell is how I spend my day...

@DavidCarlisle hopefully we will have good news in v7 :)

@PauloCereda I may have switched to wsl by then!

@DavidCarlisle we could go for Office too

Michael Palin on BBC talking about Prince Philip

@DavidCarlisle @JosephWright ^^

@DavidCarlisle yes, exactly :)

@michal.h21 ^^ another test passing, thanks configured as you suggested (@JosephWright)

@DavidCarlisle ooh

@DavidCarlisle great! does anything else fail?

@michal.h21 the one with \directlua in a math expression, that fails as it uses mathjax configuration so gets passed verbatim. If it uses mathml,mathjax I think it would expand it first so it might work, but otherwise documenting it doesn't work is OK I think. The other one is the mathtools example, that is a problem with mathjax there is an open issue there to update the mathjax2 mathtools support to mathjax3 so I don't think it is worth doing anything to configure make4ht to avoid that

@michal.h21 so on balance make4ht does exceptionally well, far better than I expected it would when I added it.

@michal.h21 I have now added latexml and set Bruce a challenge of keeping up with you:-)

@DavidCarlisle @michal.h21 Karl removed the duplicated pgfsys-driver, but one should probably try now to get yours into pgf as it seems to work better.

Yannis has created an excellent new tag!

@DavidCarlisle carlisle

@UlrikeFischer If I can get a gold tikz-pgf badge I ought to be able to get a gold badge for that don't you think?

@DavidCarlisle [tag:Great Master David Carlisle with Gold Badge] ? (but they don't like spaces ;-(

@UlrikeFischer people who put spaces in tags deserve no sympathy

teh-great-master-david-carlisle ?

@UlrikeFischer ulrike_the_sword_wielder