TeX, LaTeX and Friends - 2019-04-06

« first day (3079 days earlier) ← previous day next day → last day (1847 days later) »

5:24 AM

@DavidCarlisle Is it okay for you if I delete our conversation in github.com/pgf-tikz/pgf/pull/647? It doesn't really have anything to do with the pull request itself.

1 hour later…

6:39 AM

@Schweinebacke I think I'll follow your suggestion and stay away for a month, let's see, when I'll be back, if I am strong enough to totally ignore bad users. See you soon! — CarLaTeX 7 mins ago

@CarLaTeX Enjoy the spring sun! Uugh, now, I have to prepare myself to the worst, because I'm the pig that makes CarLaTeX pause TSX! Maybe I should choose to do my own next pause much earlier than expected.

6:50 AM

@Schweinebacke No, please IT IS NOT YOUR FAULT, bye! <3

@CarLaTeX This may sound stupid but.. what are you doing/going to do?

7:16 AM

@UlrikeFischer I'll fix this: just a question of looking for the unicode table I think

7:32 AM

@JosephWright why is expl3 accessing the table? Which functions are needed from it?

@UlrikeFischer Needed for charcat, needed for \char_generate:nn

@UlrikeFischer I've nearly sorted it

@JosephWright then what will you do if the table is not available?

@HenriMenke I was about to comment ... I will here instead :)

@UlrikeFischer Use the one from l-unicode.lua

@UlrikeFischer utf.char should be a replacement: there does seems to be a migration path

@JosephWright I'm not sure but I think there were other packages using it. And it is docunentated in the luatex manual so even if expl3 can do without it I should probably keep it.

@UlrikeFischer Sure, that's a different question

@UlrikeFischer Basically, I think local local_utf = utf or unicode.utf8 is about right ... but I'm not doing that

@UlrikeFischer Fixed :)

@UlrikeFischer I guess I should release to hit the TL'19 DVD

@HenriMenke TL will never drop tex as TeX90: for stability this is vital

7:47 AM

@JosephWright I was musing to myself; -) it is a bit of a nuisance to have to edit the file at every pull. But I don't want context lets decide about the library.

@HenriMenke In the end, whether to support use of pgf without e-TeX is down to you; it's not a bug to require e-TeX provided you say you will. As @DavidCarlisle says, the number of people using TeX90 for new material is vanishingly small. My worry would actually be how many registers you need ...

@Schweinebacke hi mr. pig!

8:14 AM

@HenriMenke I wouldn't worry about non-etex users. A few will complain a bit and then adjust. But you should worry about developers: If you are forcing them to keep etex primitives out of their code, that makes coding more difficult.

8:32 AM

@UlrikeFischer Like I said, it's a question of telling people where the line is. For developers, that should really not be an issue: it just has to be clear. For example, one could add it to the README

9:26 AM

@JosephWright For me it would be an issue. I got invited to tikz/pgf too and now assume that I want to fix a bug or add some feature which could be done easily with etex but not in tex. Why should I bother to spent my time to find and test work-arounds? After some time existing tools should be allowed to use.

@UlrikeFischer Well yes, but you could just turn down the invite. It's the same as any tool, like I said: there's a question of where the line is on what features to use. pgf I guess goes back a long way, and the policy has always been the same.

@UlrikeFischer One could argue that for stability the policy can't change: pgf has always worked with TeX90, there will be files out there using it with TeX90, probably maths papers. And the whole point of using TeX90 is archival stability.

@HenriMenke sure (I deleted mine) It is also of course your call to allow e-tex or not, but long term (or even short term) I can't see how that is a sustainable position. The number of people using plain tex with dvi output and need compatibility with pre 1990 engines while using tikz must be in single figures at most...

@JosephWright people caring for archival stability shouldn't update pgf ;-) And regarding "adjusting to changes": Did you see the nice answer regarding the \smash change?

@UlrikeFischer Ah, back with the whole 'how to deal with updates' business: that's fine of course until you start worrying about accumulated bugs/new features

10:00 AM

@JosephWright sure but if you want repairs to be done you have to accept that people doing the repair want to use more comfortable tools. And if they really want to stick to tex90 than they could adapt the code themselves, after all it is all open.

1 hour later…

11:28 AM

After I learned that git uses >>> and <<< for git ls-files -u I'm in search for a new fold mark. Do you guys know of any potential conflict for >>= and =<< (in any programming language or git or ...)? Does someone have a better looking idea?

@Skillmon \begin{macrocode} ... \end{macrocode} is I suppose what most latex code uses here...

@DavidCarlisle I don't want \begin{macrocode}...\end{macrocode} in my C code, sorry :)

@Skillmon you should, literate programming is the way of the future, don't you know....

12:12 PM

@DavidCarlisle While I generally like the idea, I dislike that I would put \begin{macrocode} on every fold (if I go for fully manual folding)...

1:00 PM

@Skillmon probably a vim thing, I have no sympathy.

2 hours later…

2:42 PM

@JosephWright Aside, but always wanted to ask: where does the name TeX90 come from? As far as I can tell from all the sources, Knuth used "TeX82" in the early days to distinguish from the earlier SAIL program (TeX78), then later just calls it "TeX".

And looking at the version history of the current TeX (say top of ctan.math.utah.edu/ctan/tex-archive/systems/knuth/dist/tex/…) I guess one could say that something special happened in 1990 because it moved from version 2.993 to 3.0 (the major 8-bit changes happening in 2.992 in 1989), but really it's a continuous evolution of the same program…

@DavidCarlisle I just like my code to be organized and while the automatic folders do a decent job, I often find myself wanting manual control, hence manual markers are required.

@ShreevatsaR TeX 3 was when Knuth said TeX is finished. Everything that follows are just bug fixes.

3:18 PM

Does anybody know why pdftexcmds defines \pdf@strcmp as \def\pdf@strcmp#{\pdfstrcmp} if one uses pdfLaTeX but in XeLaTeX it does \def\pdf@strcmp{\strcmp}?

3:37 PM

@Skillmon because in xetex \pdfstrcmp doesn't exist but \strcmp does? or what do you mean?

@ShreevatsaR 8-bit TeX Released in 1990, has some different primitives to TeX82

@Skillmon Yes :)

@Skillmon It's arguably a mess-up :)

@Skillmon BTW, you might want to look at the expl3 implementation: for LuaTeX there are some subtlies

@UlrikeFischer yes but is the #{ needed in one case and not the other? we should check...

@Skillmon From TeX by Topic:

@DavidCarlisle ah, no idea ask @JosephWright ;-)

>There is a noticeable difference in the treatment of hyphenated fragments between TEX2 and TEX3. TEX2 insists that the part before the break should be at least two characters, and
the part after the break three characters, long. Typographically this is a sound decision: this
way there are no two-character pieces of a word stranded at the end or beginning of the
line. Both before and after the break there are at least three characters.

@DavidCarlisle They should both have it, I guess

@UlrikeFischer Fix for ConTeXt and expl3 sent to CTAN

@ShreevatsaR I'd say with a modern hat on 2.992 looks like '3.0-beta-1' ...

3:45 PM

@JosephWright why not leave it up to the underlying primitive to pick up the <general text> ? not sure I see what #{ is doing other than error cases?

@JosephWright good. How do I tell l3build install to put the stuff in TEXMFHOME and not in the current git folder?

@DavidCarlisle I think because \pdf@strcmp is defined as a macro with two arguments thus should not accept anything before the first {, which the primitive presumably allows (have to check...)

@DavidCarlisle:

\pdfstrcmp\relax{abc}{abc}
\bye

@UlrikeFischer It should do as-standard: do you have an odd setting for TEXMFHOME?

@UlrikeFischer The install target simply queries TEXMFHOME unless --texmfhome is given

@DavidCarlisle I'm pretty sure its Heiko being extra-careful

@JosephWright not sure I agree, the \relax is part of <general text> so why disallow it, using #{ doesn't do anything about the second \relax in

\pdfstrcmp\relax{abc}\relax{abc}
\bye

@DavidCarlisle Hmm, could do with an auxiliary then ...

@DavidCarlisle Perhaps Heiko missed that!

@JosephWright I think I'd rather just use the xetex simple macro definition in both case so the package is just handling the name change, not trying to suppress primitive behaviour

3:54 PM

@DavidCarlisle Fair

@JosephWright texmfhome and me don't get along well ;-). I have a kpsewhich setting pointing to a non-existing folder (which is ignored by l3build) and no environment variable. Setting the environment variable worked.

@JosephWright Good afternoon. With such kindness and courtesy can I have the opportunity to write in private in a room? I don't remember doing one in two and a half years. Could you help me? Thank you very much.

@yo' Hi, :-)

user image

5

@AlanMunn :-) :-)

@UlrikeFischer We query kpse for the value:

local function gethome()
  set_program("latex")
  return abspath(options["texmfhome"] or var_value("TEXMFHOME"))
end

@Sebastiano I can open a ''private' room but any one with mod privilege can enter them

3:59 PM

@AlanMunn LOL

@AlanMunn I prefer herbal teas, but it's delicious tea of the same brand but vanilla.

And on a related not, the brilliant James Acaster:

Brexit - Tea bag analogy

@DavidCarlisle You going to change that then?

@JosephWright I accepted, but he diverted me to another room. He tells me after giving OK access denied. What do I have to do?

@Sebastiano I would say that vanilla is to tea for me what pineapple is to pizza for you.

4:02 PM

@JosephWright I was going to suggest @Skillmon open an issue at ho-tex, haven't actually looked to see whether there was any documentation about the #

@Sebastiano I've granted you access too chat.stackexchange.com/rooms/92083/… I hope!

@AlanMunn delicious and tasty?

@DavidCarlisle Ah, right: like I said, I can only imagine Heiko was being restrictive

@JosephWright I agree about 3.0-beta-1 :) But given version 1, 2, 3 were released in 1983, 1986, 1990, seems an interesting decision to give each of them a different name (instead of TeX/TeX82 v1, v2, v3), something like Windows 95, 98, 2000... was just wondering if anyone else has used that name (looks like yes: tex.stackexchange.com/search?q=tex90 )

not important :)

Hi mr. velociraptor!

4:04 PM

@ShreevatsaR I've always called the releases 'TeX78', 'TeX82' and 'TeX90', though all of that predates my involvement so ....

@JosephWright Named in the good old Fortran tradition.

@AlanMunn Yeah, for some reason version strings are very complicated

@AlanMunn looking forward to Fortran 2020?

Browser user agents: webaim.org/blog/user-agent-string-history

2

@DavidCarlisle I suspect any new features would be of limited use for my needs. (Apart from the fact that I haven't written any Fortran code in decades.)

4:10 PM

@DavidCarlisle I didn't found any. But he uses the same command also for other primitives. Would the # make sense if a primitive has a keyword like file or so before the general text?

@UlrikeFischer you could use #1#{ in that case, but #{ just makes it an error if there is not a following {

@DavidCarlisle My main worry these days is making sure my version of PsychoPy (a Python based experiment platform) will run my experiments. Sometimes this involves successive downgrades until you find a version that works. It's all very scientific. :)

@JosephWright the problem was that the root folder didn't exist.

@UlrikeFischer Ah, that would do it

@AlanMunn ooh

4:21 PM

@AlanMunn It is known that you need to choose very careful wording to make PsychoPy cooperate. No exclamation marks. Start every sentence: "Could you consider .... ?" instead of "Do that for me".

@JosephWright I'm trying to find a way to get various experimental code in the various git folders to work together. currently it is a mess. tagpdf doesn't find the pdfresources or the wrong one.

@UlrikeFischer URgh

@marmot If only it were that simple. :)

@JosephWright does l3build check use texmfhome?

@UlrikeFischer No, everything is done in a local folder, although of course you may have searching turned on

4:43 PM

@JosephWright searching is on, and tagpdf finds pdfresources in the texmfhome. So I guess it now can build the docu if needed ...

@UlrikeFischer Great

@UlrikeFischer I'll look at the PDF stuff this evening

@JosephWright I wonder if you will understand anything ;-) some of this hyperref stuff is messy - finding out which of the hyperref commands are doing what is not easy, and they have all quite similar names ... But I think I can now add links to the toc too. And probably cites will work soon also. We are in the theater this evening, so I won't do much more now.

@DavidCarlisle the same error would be thrown for \def\pdf@strcmp{\pdfstrcmp}.

5:11 PM

Beautiful. Thank you very much for taking the time to formulate this detailed answer, and for thinking ahead so many steps! — Zubo 4 mins ago

@Zubo What you do not know: I've made an answer. After trying to send it, I've found out, that I've forgotten to login an. After login, the ¾ of the answer was removed. So II've written an new answer.

@Zubo But is was a good occasion to improve the first suggestion. ;-)

@DavidCarlisle been there, done that

@Skillmon no, that would allow \relax same as the primitive

@DavidCarlisle yes, \relax would be the exception, you're right.

@Skillmon the only reason Heiko could have had for using the #{ form is to not allow \relax but as I said to Joseph above it doesn't really work as you can use {aaa}\relax{bbb} with \relax before the second argument, so I think it would be better not to use it at all, as in the xetex version

@DavidCarlisle Or to use \def\pdf@strcmp#1#2{\pdfstrcmp{#1}{#2}}.

@DavidCarlisle allowing a single token without a brace to be an argument.

5:26 PM

@Skillmon personally I wouldn't do any of these, just \let\strcmp\pdfstrcmp in pdftex, define the lua version for luatex for that name, and use the primative as-is in xetex:-) but I can't make the package do that...

@DavidCarlisle personally I'd \let those, too. Don't you have write permission for his git?

@Skillmon I set the git up:-)

@DavidCarlisle But then the expansion steps vary

@JosephWright that's an argument...

@JosephWright not really since you can make them all expand in one step

5:28 PM

@DavidCarlisle sooooo, why can't you make the package behave that way? Are you only allowed to fix some issues by Heiko?

@DavidCarlisle Not the LuaTeX one

@Skillmon because it would be a breaking change and not in the spirit of the package, all the comands are defined to be macros not let to primitives in this package.

@DavidCarlisle then change it to \def\pdf@strcmp{\pdfstrcmp}, works for me.

@JosephWright why not? can't all the work be kept in lua, ooh of course you need one macro expansion to get the \directlua, yes..

@DavidCarlisle :)

5:31 PM

@DavidCarlisle that would be a reasonable addition to Lua, something like macro.new.directlua(\pdf@strcmp,<stuff>) such that \pdf@strcmp already implicitly contain the \directlua and only needs one expansion.

@Skillmon I was thinking same, except luatex is abandoned...

@DavidCarlisle yes/no, as far as I understood the current version is frozen and they'll come back with a V2.

@Skillmon but the new engine isn't likely to be able to run cross platform formats like latex, as far as I can see

@DavidCarlisle that's a really stupid design choice, imho.

@DavidCarlisle remember the days when we thought luatex would be the future?

@Skillmon Not from Hans' POV: LuaMetaTeX is for ConTeXt ...

5:34 PM

@Skillmon not necessarily

@Skillmon Well it is, in the sense that ConTeXt is the future for Hans ;)

@Skillmon it still might be.

@Skillmon Probably it is, for several reasons

@JosephWright well, I thought, we thought it will be the perfect engine for everything once it's stable enough and stuff.

@Skillmon More or less the situation now ...

5:36 PM

@Skillmon if the issues surrounding using harfbuzz from luatex (or luametatex) are fixed then xetex is really history, and it's hard to justify an 8bit system for text this century so pdftex ought to be history, so the need for cross platform formats is not clearly always going to be needed.

@JosephWright and where is my macro.new.directlua?

@DavidCarlisle Exactly

@Skillmon Ah, well that is a thing

@DavidCarlisle will it be able to run a luaLaTeX format?

@Skillmon luametatex not at present ...

@JosephWright so it can only run Context?

5:41 PM

@Skillmon Yes: not even plain yet

@Skillmon I've asked Hans ...

@Skillmon that depends on your definition. If we gave up making the same code running on pdftex/xetex/luatex and just targetted luametatex then we could get a latex v? working certainly, but if trying to keep it running on pdftex, then probably not

@JosephWright but why? I mean I have no idea of the architecture of neither LuaTeX nor LuaMetaTeX, but shouldn't it be starting with implementing the primitives?

@Skillmon no. Not if you view it as "design a lua and unicode based typesetting system inspired by tex" you would not start from implementing all tex82's quirks

@Skillmon Er, well, yes, except I can't even get it to read a source file due to a Lua callback issue. There must be something very early in the format stuff. I could/should read the log from building ConTeXt and track it down

@Skillmon it only runs on context as currently there is no docuemntation and no source so only Hans can make a format that works with it, but that's just because it's a first preview release

5:44 PM

@DavidCarlisle So I should stop coding macros and stuff for (La)TeX and just wait till LuaMetaTeX is the thing, relearn everything for it and start anew?

@Skillmon some people asked that about latex3 in 1993 too, I guess most of them stopped waiting and coded for 2e. It is hard to know...

@Skillmon There are still a lot of macros in ConTeXt

@JosephWright yes, but I have no idea of Context. At the same time I consider myself quite knowledgeable in LaTeX.

@Skillmon :)

@DavidCarlisle @JosephWright Do you feel it is a bad thing that they "leave" luatex and start this luametatex thing? Or is it too early to say?

5:48 PM

@Skillmon you and millions of others so one way or another some system will be provided that keeps things working, with possibly some changes, but there are costs to supporting the underlying code base running on different tex engines, especially if they start to diverge.

@JosephWright I know understand why there are people using TeX90

%D \module
%D   [       file=cont-en,
%D        version=1997.08.19,
%D          title=\CONTEXT,
%D       subtitle=\CONTEXT\ English Format Generation,
%D         author=Hans Hagen,
%D           date=\currentdate,
%D      copyright={PRAGMA ADE \& \CONTEXT\ Development Team}]
%C
%C This module is part of the \CONTEXT\ macro||package and is
%C therefore copyrighted by \PRAGMA. See mreadme.pdf for
%C details.

\catcode`\{=1 \catcode`\}=2

\def\defaultinterface  {english}
\def\defaultlanguagetag{en}

@DavidCarlisle ^^^ I can understand this part :)

@JosephWright I guess there are bigger differences elsewhere:-)

@JosephWright I don't. What is %D and what is %C?

@mickep Well we'd prefer someone was looking after LuaTeX, but on the other hand it is now in C (well, CWEB), so probably a 'helper' can be found. Contrast pdfTeX/XeTeX, written in WEB

@Skillmon They are comments: ConTeXt uses something a bit like DocStrip, but not exactly, so they are 'magic' information

5:50 PM

@JosephWright I guessed so, but what exactly is %D and what exactly is %C?

@JosephWright as long as you can't tell, you don't fully understand that above :)

@JosephWright I see. As a simple user I prefer the syntax of ConTeXt (and write my own documents using it), but I still must be able to write LaTeX for research (math). I don't know much about this new thing, but I certainly hope it will lead to some good things for users in the end.

@DavidCarlisle Point being, that is loaded first, so there must be a place later where the Lua issue comes up

@mickep Fair enough: I've stuck to LaTeX as I prefer the font handling and also the environment syntax, but this is personal preference

@JosephWright yep,

@JosephWright Personal preference, indeed. I like that start/stop syntax and the in my opinion more coherent syntax with key=value and so on. Math works better in LaTeX, though, and that is sometimes a pain (I just wrote a book using ConTeXt, and there are still some small things I'm not 100% happy with).

@mickep bad for who?, it makes perfect sense for context, what it means for having a supported unicode-aware tex-compatible system in the future, it is hard to say at this time.

5:55 PM

@DavidCarlisle Good question. The community, luatex, users, ... Perhaps with emphasize on the development of luatex (which I was happy about).

@mickep Oh, not one is making any bones about keyval being better than a lot of place-based arguments: it's a question of history there (ConTeXt bit that bullet at teh word go, I guess at the time with really poor performance)

@JosephWright outsourcing the list handling to Lua wouldn't be that bad of a performance impact would it? One might think that stripping spaces is faster in Lua than in TeX, as well as list handling, since Lua has a list type...

@DavidCarlisle I suspect luat-cod.lua is where the issue is: it's loaded early

@JosephWright I haven't grabbed the new code at all yet

@Skillmon I thought about that once ... but Lua has no concept of catcode, so if you have anything passed with non-standard catcodes you loose that entirely when you pass back to TeX

@DavidCarlisle Ah, well I think the .log is the same more or less as ConTeXt on LuaTeX

5:59 PM

@JosephWright ohh

@Skillmon that comes back to what we just said about cross engine code. You could define some argument list processing handling in lua but hard to make it act like a macro based one. If (like context) you only target one engine then you can make a choice. If you need to keep more or less compatible code running in xetex then there is no choice.

@Skillmon I did give it quite a bit of thought ...

@DavidCarlisle I was speaking about Context and it's performance issues in the early days there.

@Skillmon That was all done in pdfTeX (MkII)

@Skillmon context early days had no Lua

6:01 PM

@Skillmon The performance since LuaTeX was developed has been fine: I was thinking about 1992

@JosephWright since when did they start to use LuaTeX

@JosephWright that's a while...

@Skillmon since they implemented it:-)

@Skillmon About 12 years I think

@Skillmon Sure ...

@JosephWright if you worried about context performance in 1992 you clearly never tried a 1992 latex3 :-)

@DavidCarlisle :)

@DavidCarlisle ConTeXt released around 1990?

6:06 PM

@DavidCarlisle how big/well developed was L3 back then?

@JosephWright about then I guess, possibly a bit later (for the public versions)

@Skillmon there was a self standing latex3 format (where you had to use \zzz_zz:xxx as all the primitives were undefined), when I joined the project in 1992, I had access to a Sun microsystem at the time so it was almost usable for small documents, but this was not like expl3 it was a fully incompatible format so no existing documents or style files worked with it

@DavidCarlisle I thought expl3 is only an interim solution and eventually L3 will again be incompatible?!

@Skillmon possibly (or possibly not)

Time for dinner. Good bye for now.

@Skillmon duck?

6:12 PM

@Skillmon Er ...

@Skillmon Well if we ever get a totally stand-alone format with all of the toys we want, then yes other than the user syntax. But that seems extremely hard to pull off.

@DavidCarlisle Hmm, must be in the context script itself or something: I just tried luametatex --ini cont-en.mkiv, and I got the same issue as I did with latex.ltx

6:34 PM

@DavidCarlisle sweet potato with kidney beans and mango.

@Skillmon mango, @CarLaTeX would have recommended pineapple, but I suppose it's OK as a substitute.

@DavidCarlisle did taste great. My wife wanted to try something new.

I was just reading an old Leslie Lamport interview, and wondered if I had commented about it here. I had. 5 years ago. How time flies.

Jan 22 '14 at 22:16, by Faheem Mitha

Leslie Lamport interview:

Slightly more than 5 years ago, actually.

3 hours later…

9:13 PM

@egreg Let me know if you've read all my comments that I'm removing everything immediately. I hope in a friendly conversation in a private chat room where I discussed with Joseph.

9:49 PM

@Sebastiano Yes, I read them

@DavidCarlisle, @barbarabeeton, @egreg, @UlrikeFischer (others) seen the TUG list?

@JosephWright yes, just read it.

@JosephWright Boris' post?

@egreg No, Robin's

@UlrikeFischer Reading it now in detail ...

@JosephWright not yet

9:55 PM

@JosephWright Where is it?

@JosephWright Is it the news that DEK will be at TUG?

@egreg No idea!

tug.org/tug2019

@JosephWright sad (but not surprising) she feels personally about JF's constant harassment.

10:39 PM

@egreg There is a little typo in your answer tex.stackexchange.com/a/483596/132405 (you wrote "I suggest using the task package." but it's the tasks(with "s") package, as you used the correct name in the code)

@quark67 Thanks, fixed!

@egreg if only all your other answers could be fixed so easily:-)

@JosephWright -- thanks for heads up. got it in "regular" email (i'm subscribed to the list).

11:06 PM

@egreg TeXstatic? Really?

« first day (3079 days earlier) ← previous day next day → last day (1847 days later) »

Transcript for

Apr '196

$TeX - LaTeX$ TeX, LaTeX and Friends

\begin{chat}

join this room
about this room

00:00

06:00

12:00

18:00

all times are UTC

$TeX - LaTeX$

site design / logo © 2024 Stack Exchange Inc; legal

mobile