TeX, LaTeX and Friends - 2018-10-30

« first day (2921 days earlier) ← previous day next day → last day (2003 days later) »

5:40 AM

@barbarabeeton Thanks very much for the link to the history of computing articles... I look forward to reading them when they are published :-)

@KhaledHosny I scrolled up and read your comments on how fonts are handled in XeTeX / LuaTeX; I agree and thanks for the work :-) If you'd like to test with some Indic scripts, I just posted some minimal examples at @UlrikeFischer's question: tex.stackexchange.com/a/457459/48 (though these are trivial and no doubt HarfBuzz will handle them correctly... the point is just that LuaTeX doesn't)

3 hours later…

8:48 AM

@UlrikeFischer Thanks. Indeed putting a manual \par before closing the group fixes the issue, but it does not explain the difference that results from copying the node list one by one, I’m sure something gets lost here and I’d like to know what is it.

@ShreevatsaR Thanks. If you have more examples of scripts currently broken with LuaTeX that would be helpful. I’ll start doing extensive testing once I fix the major remaining issue in my code and find a way to integrate with luaotfload.

9:15 AM

'ello, Psmith is now fully operational.

/texfortune

Your fortune cookie: LuaTeX drivers are funny.

@KhaledHosny I'm not sure that I did understand what you are doing, but imho if you start code has \pardir in a group without ending \par it is faulty and so bad things can simply happen ;-).

@UlrikeFischer The problem is not the \pardir, moving it out of the group makes no difference, the problem is with the \textdir and I’d expect that to work here, but that is orthogonal with the other problem. When recreating a node list by copying the nodes and inserting them into a new list causes any difference here.

/texfortune

9:30 AM

Your fortune cookie: Tricky area.

hmmmm

/texfortune

Your fortune cookie: We eat animals, not make them team members.

@KhaledHosny then you should make an example without the \pardir in the group.

oh

@Psmith hey. Don't frighten the bear again.

9:31 AM

@UlrikeFischer 'ello, I bid you good day! Type /help to see all my commands.

/texfortune

Your fortune cookie: Standard team policy, always blame someone else.

@DavidCarlisle ^^

/blame

@DavidCarlisle Let us blame the dog

9:33 AM

@PauloCereda oh for once it did blame someone else

@DavidCarlisle 'ello, I bid you good day! Type /help to see all my commands.

@DavidCarlisle ooh

ooh

/texfortune

Your fortune cookie: XeTeX is a palindrome. Wow.

9:37 AM

/texfortune

Your fortune cookie: This sounds like a task for Bruno.

/shutdown

Initiating shutdown sequence. Cheerio!

@UlrikeFischer I just did that.

9:51 AM

@DavidCarlisle regarding the oberdiek tests: I didn't understand yet your problem with the dtx and couldn't find a script or something like this that runs the original (or other tests). Can you show an example? I would like to avoid to change everything to lvt and then it doesn't work there either ...

1 hour later…

11:03 AM

@JosephWright you don't feel an urge to rewrite a package in expl3? ;-) The embedfile package needs a patch to add the /AFRelationsship key to the Filespec dictionary and the longer I look at the package the more I think it could do with less \expandafter, \toks and with a few carefully named catcode tables. And I would also drop all the plain tex loading stuff.

@UlrikeFischer Sounds good

@UlrikeFischer You remind me I want to finalise the catcode table support stuff

@JosephWright well rewriting a oberdiek package to use them is probably a very good proof of concept - there are catcode changes and restoring everywhere.

@UlrikeFischer Yup

@UlrikeFischer OK, now I've got benchmarking out, and Lua ID nearly sorted, I'll try to finish that next

'ello, Psmith is now fully operational.

ooh another duck is here

11:07 AM

ooh (6 hits so far)

ooh (7 hits so far)

ooh (8 hits so far)

ooh (9 hits so far)

ooh (10 hits so far)

OH NO

/shutdown

ooh (11 hits so far)

Initiating shutdown sequence. Cheerio!

@UlrikeFischer We just need to decide on what makes sense as an interface: I'll lay out the options (again) on the team list

'ello, Psmith is now fully operational.

There we go again.

ooh a new bot

11:10 AM

ooh (12 hits so far)

/texfortune

Your fortune cookie: iPad Mittelbach

2

Oopsie

/texfortune

Your fortune cookie: This sounds like a task for Bruno.

@UlrikeFischer Were you happy with the interface I proposed? (@egreg, @DavidCarlisle views also welcome)

@UlrikeFischer You know that my job is more-or-less to rewrite everything in expl3? ;)

11:20 AM

@JosephWright I wouldn't use begin/end, this indicates grouping. save or store to save the current state and restore or load to load one sound best to me.

@JosephWright that's why I asked ;-).

/texfortune

Your fortune cookie: Tricky area.

hmm

/texfortune

Your fortune cookie: Do not use, unless you are on the team.

/texfortune

11:31 AM

Your fortune cookie: Catcode everything.

@UlrikeFischer Hmmm, OK: might also be relevant to benchmarking (current tic/toc don't group)

11:48 AM

@JosephWright I'm no consequent. In tagpdf I use begin/end too, and the commands don't add groups either. ;-(

@UlrikeFischer Hmm: I did think we were just using begin/end generally for 'matched stuff'

@JosephWright then it is okay in tagpdf. But imho for catcode tables it still won't fit, as there is no need to match. You can always store a table and never restore it.

@UlrikeFischer Yeah, that was something I was pondering

@JosephWright one can spend more time finding the correct command name than writing the code ;-)

hi

I needed help with creating a few new templates for things like CV, class notes (I plan to Latex-ify all the notes delivered in the classes I attend at the university), and some other mathematical notes I prepare

but not sure where to find a good and easily accessible and easy to understand guide to it

11:58 AM

/texfortune

Your fortune cookie: Catcode everything.

@adityaguharoy Do you use LaTeX on a daily basis?

yes to create mathematics notes

not very regularly, but whatever paper or mathematics document I prepare these days are created using Latex. And I am a very active user of a forum (aops) where I write mathematics using Latex regularly

@adityaguharoy Perhaps it is a good idea to find a document that looks like what you want to achieve, than start constructing the elements based on the underlying chosen document class and potential packages. :)

yes, but how can we find the code that was written to create the document

?

12:05 PM

@adityaguharoy As the author :)

@adityaguharoy I didn't say that, I suggested to mimic the visual. That's what we usually do. :)

no I mean I won't do that, I mean I wanted to learn from some example codes

I mean something that are free to be referred to as examples to learn the process

@adityaguharoy So ask the author, as Joseph suggested.

gotcha, thanks

@adityaguharoy There is code freely available in some websites, as well as in the threads in here. It's a good spelunking adventure. :)

12:09 PM

can you give some links @PauloCereda

@adityaguharoy texample.net

thanks

/texfortune

Your fortune cookie: iPad Mittelbach

hm

/texfortune

12:16 PM

Your fortune cookie: Blame Joseph, it is for the best.

3

ooh

ooh (13 hits so far)

@JosephWright ^^

/say

Alas! no, said Psmith, the pet of our office, waiting for the first to be done then, he said, to bring gats into der grill-room vorbidden

wat

12:37 PM

@UlrikeFischer That was one of the things that had me pondering about interfaces ...

1:28 PM

@JosephWright and just to make your decision more difficult: there is also "dump" for the save part.

1:39 PM

/texfortune

Your fortune cookie: Standard team policy, always blame someone else.

4

2:10 PM

/say

Are you Bat Jarvis to speak, deprived of its strength on the chest

2:54 PM

> Tomorrow is Halloween, but this week at the Fedora Project, we're not scared of anything. Today, we officially release Fedora 29!

/shutdown

Initiating shutdown sequence. Cheerio!

3:15 PM

@CarLaTeX hii, what's up?

user image

@CarLaTeX ^^^^^ lol, are you strict with that? So all muy messages include "Thanks!!* and " Is that possible?"..... you should edit it too

3:33 PM

48

Q: Why is "thanks" inappropriate at the end of a question?

This question seems to indicate that it is inappropriate to say "thanks" at the end of a question, but does not provide justification. In a question I once asked, such thanks were removed and I tried to get justification, but to no avail (see the comments below the question). Is it inappropriate...

discussion etiquette questions

3:47 PM

@manooooh I re-edited for other reasons, and with the occasion I leave out the "thank you" sentence. I don't edit only for that :)

4:06 PM

@DavidCarlisle I am clearly disagree with that

@CarLaTeX ok

@manooooh that's the thing about being in a community, there are rules and conventions even if you don't agree with them all. search for "thanks" in the meta site, you will see multiple discussions on that.

@DavidCarlisle we do not are robots

@manooooh no but humans should follow rules too, here are the instructions given to new users, which already tell you about this

172

Q: Welcome to TeX.SX!

Welcome to TeX.SX, the free, community driven Questions and Answers site for users of TeX, LaTeX, ConTeXt, and related typesetting systems. Here is some information to help you get started and to make your work with this site a pleasant experience for you and all other users. This site is for ...

support editing new-users

@manooooh rules exist exactly for scenarios of disagreement. :)

@DavidCarlisle there are a big difference between "OMG THX YOU YOU SAVED MY LIFE" and the people that comment including a reason of why is the best answer. Or "Thank you so much for your huge help" vs. "Thanks!"

4:12 PM

@manooooh The network policy is pretty clear: focussed Q&A means minimising 'noise', and that includes 'Thanks'

@JosephWright again, I am totally disagree with that

@manooooh it's fine to disagree with it (I am not sure I agree with it either) but you should follow the convention anyway.

@CarLaTeX already did the formal work pretty well, but the same rules must be applied to all users, including me

@manooooh exactly, so you should expect to see "thanks" removed if your questions are edited (but typically they are not edited just for that as minimal edits are discouraged)

@manooooh I agree with @DavidCarlisle (omg!)

4:18 PM

@DavidCarlisle so suppose that now any question follows exactly the rules, except for "Thanks" at the end. Would you edit all the questions now?.... Or change that rule? seems more realistic the second one, no?

@manooooh no as I said (and as @CarLaTeX) said minimal edits are discouraged so i wouldn't edit just for that, especially for new users but you have been on the site long enough that you could be expected to follow the conventions without relying on others to edit later

@CarLaTeX so you agree that Hawaiian pizza is good?

@DavidCarlisle some users are here long enough and faulting to the rules... so that is not applied here

@UlrikeFischer There is no script that runs the existing tests, and they are not run.

I fault the rules and been here more than 100 days, for example

@manooooh well don't.

@UlrikeFischer the problem with updating the dtx files is the repeated version strings, take pdftexcmds.dtx for example that has the current version 0.29 in 9 places. 8 of which need to be changed to 0.30 if it is updated along with adding a new 0.30 entry in the change log within the file. It was a pain when I adopted these but I have some mechanised emacs support now so it's not too bad but...

4:27 PM

@DavidCarlisle No, :P

@CarLaTeX OK, you'll have to stick with @PauloCereda and have nutella pizza then. (That does sound pretty bad)

@DavidCarlisle Nutella pizza is not a pizza

@CarLaTeX vim users have no taste so it's Ok for them I suppose.

@DavidCarlisle @PauloCereda ^^^

4:42 PM

@DavidCarlisle Nutella pizza is good

@CarLaTeX boo

4:56 PM

@PauloCereda You should create a bot command for comments on editors

5:25 PM

@CarLaTeX That happens also to me, once in a year.

5:59 PM

@egreg lol

@CarLaTeX what's to laugh about?

@DavidCarlisle Agreeing with you is funny :)

@CarLaTeX I do it all the time.

@DavidCarlisle The contrary should be difficult :)

Wanna play Pinturillo?

6:46 PM

@manooooh no time now, sorry

2 hours later…

8:22 PM

hello

\usepackage[english,arabic,french]{babel}

it. have a sens to write this?

@Vrouvrou ? it means you load commands for English, Arabic and French (French being the document default)

i have to write something with arabic

and a code work only with \usepakage[French]{babel}

when I put the three

@Vrouvrou you would be best to use xetex or luatex

no it didn't work with xelatex

@Vrouvrou "it didn't work" is not a useful description. what did you do and what error did you get.

8:27 PM

! LaTeX Error: Command \fa unavailable in encoding T1.

@DavidCarlisle probably he had some local script/list.

@Vrouvrou that is not surprising, T1 is the encoding for westen european latin scripts,

@UlrikeFischer I think he had a whole system for building the dtx from some other source (which is why they are so hard to edit manually) but I just picked up what was on ctan and tried to reverse engineer some kind of build system:-)

@DavidCarlisle what I must put ?

@Vrouvrou you have been on the site long enough to know that it is much easier to help if you ask a proper question with a proper example on the main site.

@DavidCarlisle I see. And one can't go over all the files as they all have different versions, even l3build tagging isn't easy here. Tricky.

8:32 PM

@UlrikeFischer I suspect the first step should be that I pull all the test files out of the dtx so they are not generated and check them in as tex source and then we try to convert them to lvt one by one....

the problem is that the tex.files is too long I don't know what I must delete

@DavidCarlisle Sounds like a plan to me

@Vrouvrou just make a test file with one sentence in arabic and one in french, if that works good, if not, ask a question posting the test file to the main site

@Vrouvrou for example here is an english/arabic babel example (first hit for "arabic babel" in the site search

1

Q: babel arabic changes enumeration level

I am trying to add a few short words of Arabic to my English document. This is working, by and large: \documentclass[fontsize=12pt,paper=a4]{scrartcl} \usepackage[LAE,T1]{fontenc} \usepackage[utf8]{inputenc} \usepackage[arabic,english]{babel} \begin{document} \section{Introduction} It is call...

numbering koma-script babel arabic

@DavidCarlisle I have already around 30 lvt, and about 170 to do.

@UlrikeFischer is there no bulk edit we can do to get a bunch of them converted? (I must admit I haven't looked at them recent;y)

@UlrikeFischer I already simplified the build a lot, originally it required xslt to convert an xml readme.xml and changelog.xml to tex then make a pdf README.pdf and CHANGELOG.pdf. I like xml and xslt but....

8:41 PM

@DavidCarlisle quite a lot (normally one per package) seem simply to check the catcode states in and after the package. Not sure if they are really needed, but adding them is easy. For the rest one has to really to look into the files and try if they run and if engine specific tlg are needed. But I had already three failures which needs more investigation.

@UlrikeFischer yes that's really why I never yet added running them to the build script as I had no sense of what was a passed test (other than not giving an error) or for that matter I didn't have a good list of what tests there were as they just get made during the (very long) build process.

@DavidCarlisle I didn't try to understand everything but imho "not giving error" is the main point.

@UlrikeFischer yes, which means I suspect that it's an awful lot of code for what are in fact very light tests :(

@DavidCarlisle no, some of them looks very sophisticated and test quite a lot. But the conditions are not checked by log comparision as with l3build but by either writing ok in the log or giving an error.

@UlrikeFischer wher edid unexpded go... (./soulutf8-test4.tex
LaTeX2e <2018-12-01> pre-release-6
! Undefined control sequence.
\@onefilewithoptions ...@a {\endgroup \unexpanded

@UlrikeFischer oh its undefined in the test, OK

@UlrikeFischer actually not OK, what do you get for pdflatex soulutf8-test4.tex ???

9:05 PM

@UlrikeFischer On the cctab business, I think as @DavidCarlisle says we are probably OK as we are, perhaps just with begin/end needing a little thought

@DavidCarlisle an error. But if latex gets modern and uses \unexpanded in the kernel undefining it is not so a great idea ...

@JosephWright it's a shame that the most common use case, saving at start of some package code and putting stuff back at the end requires maintaining a stack, but if we could afford teh bytes to maintain a stack for@ in 2e, can prpbably afford a catcodetable stack now:-)

@UlrikeFischer but haven't we had etex based formats for years, when did these tests last run...

@UlrikeFischer I blame @JosephWright for that

@PauloCereda^^

@DavidCarlisle Really most people shouldn't be worrying about that ... but to be on the safe side, I think long-term I will use the table approach for \ExplSyntax(On|Off)

@UlrikeFischer The kernel requires eTeX, so we can do that

@DavidCarlisle I think FMi said it in about 2003 ... and then sort-of forgot

@DavidCarlisle well when was unexpanded used the first time in \@onefilewithoptions?

@UlrikeFischer oh this year, blame Frank then:-)

@JosephWright no that was "requiring" I was thinking when did texlive build the formats with etex and I think that is much earlier, perhaps always

9:15 PM

@DavidCarlisle No, it's about the same time (I did check this recently: it's mentioned in the TL release notes). Something like 2002/2003 they dropped pdfelatex, etc.

@JosephWright @UlrikeFischer speaking of \@onefilewithoptions i have some changes to finish testing....

@JosephWright Oh OK

@DavidCarlisle Ah, of course

@DavidCarlisle better change the test. Why is it undefining \unexpanded anyway? Sounds very odd that it shouldn't exist when soulutf8 is loaded.

@JosephWright and as we said in the call, any re-arrangement of that code was bound to break something....

@UlrikeFischer no Idea i literally landed on that text by accident just picking on one to see how it tests success

@UlrikeFischer would it help or delay you if I checked in the oberdiek bundle with the test files as .tex files and stop them generating, then you could do a PR against that?

@DavidCarlisle I copied them simply from the doc folder and have them here github.com/u-fischer/test-oberdiek/tree/master/source (I delete files for which I have already created an lvt). So currently I'm not using the oberdiek git.

9:25 PM

@UlrikeFischer Oh OK, I may do the re-engagement then, then we can copy in the lvt files whenever you are ready.

@DavidCarlisle ok, it will take a few days. If you want to take a look: The testfiles-failed folder contains files that failed, I put comments at the start about the problem.

@UlrikeFischer I probably should do the package loading things for the next 2e release, which will be all my tex time for this week so no rush:-)

@DavidCarlisle Sounds good

@DavidCarlisle I'll do the Lua module stuff assuming we get a decision ...

9:51 PM

@DavidCarlisle even if I rush - 170 files take more than a week ;-).

@UlrikeFischer I'll check in a re-arranged setup tonight most likely but none of the test files will be really looked at, just not generated

@UlrikeFischer is it expected that require("lualibs-table") (or pretty much any other lualibs module I tried) to fail with errors like attempt to index a nil value (local 'patterns')?

10:09 PM

@KhaledHosny it doesn't surprise me, there are dependecies. Load lpeg first: \directlua{require("lualibs-lpeg")}.

@UlrikeFischer OK

And I thought Python is a bad programing language with its dynamic typing, Lua really teaches bad habits.

(or TeX)

@JosephWright @UlrikeFischer I pushed something to github, you can now go...

$ l3build check
Running checks on
  accsupp-test1 (1/193)
          --> failed

  alphalph-test1 (2/193)
          --> failed

  alphalph-test2 (3/193)

It's going to chug away for a while then tell me that I have 193 failures....

@UlrikeFischer is there a reason that lualibs-table couldn't require lpeg itself? doing areqire twice does nothing doesn't it?

10:29 PM

@DavidCarlisle there are not really meant to be loaded as standalone files. The normal call is require("lualibs"), and the documentation actually has a section about "2.2 Loading libraries Individually" and describes which modules are needed in any case.

@UlrikeFischer oh documentation:-)

Anyway, lualibs pollutes the global namespace (assuming that is what it is called in Lua) but I like my code to be contained and know what function is coming from what module.

10:53 PM

@KhaledHosny yes, we discussed this already. But there is not so much we can do about it, lualibs like luaotfload comes from context, and context has different views on this - it doesn't really have to care about other modules or packages.

@UlrikeFischer Thanks, I understand. I want to use it for a couple of small functions, but this downside outweighs the benefit of using it.

@DavidCarlisle you cheated, you simply add \input{regression-test} everywhere at the begin ;-).

@KhaledHosny but in latex it is loaded anyway, so you can use it.

@UlrikeFischer sure but a couple of the soul ones loop so I'm about to c check in a bigger cheat for them (adding \stop) but the point is it means i can test the build doesn't regenerate them and can test the build.lua does run all the tests, the tlg all say unchecked so they will all fail until we get time to look at them and check in a real tlg and better modified lvt in each case

@UlrikeFischer I do not have anything LaTeX-specific right now, I actually do all my testing with Plain TeX documents until I manage to integrate with luaotfload (or provide a compatible interface for fontspec to use, haven’t looked into it yet).

11:31 PM

@DavidCarlisle would it help if I look at the looping files first so that they can checked in? Then give me the names.

@UlrikeFischer soulutf8-test[45] they are currentlly checked in but disabled github.com/ho-tex/oberdiek/commit/…

« first day (2921 days earlier) ← previous day next day → last day (2003 days later) »

Transcript for

Oct '1830

$TeX - LaTeX$ 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