TeX, LaTeX and Friends - 2020-02-14

« first day (3393 days earlier) ← previous day next day → last day (1539 days later) »

Pablo González L

1:10 AM

@JosephWright I have already understood it, it just needs to be clarified in the documentation :)

@JosephWright Right, but, after making the changes in build.lua, the output of l3build install --full --dry-run throws:

For installation inside /home/pablo/texmf:
- source/latex/demopkg/demopkg.dtx
- source/latex/demopkg/demopkg.ins
- doc/latex/demopkg/demopkg.pdf
- doc/latex/demopkg/demo/example.pdf
- doc/latex/demopkg/CTANREADME.md
- tex/latex/demopkg/demopkg.sty
- doc/latex/demopkg/demo/example.tex

Which is exactly what I want, but, by running `l3build ctan` the `.zip` file not `example.tex`.

Actually it's a great tool, the documentation is a bit confusing for me (all the
variables defined at once) and the lack of examples doesn't help much :(

@JosephWright I added some things _by hand_ because I didn't find the right way to do them and they are useful for those of us who know NOTHING about regression tests and are
used to do things the old-fashioned way.

@JosephWright For example, I have no control over `typesetdemofiles` and as it is now my
example compiles it using the same routine from the documentation (lualatex 3
times), I tried to copy your `function typeset_demo_tasks()` routine to run
`pdflatex example.tex` but this function goes in an earlier stage than "doc" and
the file doesn't exist in that run point.

@JosephWright I can validate the input format of a tag, but I can't run tag v1.3 --date 2020-02-12 for example.

@JosephWright I would be happy to help in this example :) ...as soon as it works properly. I know you are a busy person, but if you have time take a look at the repo...I will not modify build.lua anymore. Greetings

@UlrikeFischer Dear, I could share a dummy version of the ulrikefischerdata.lua file ...I think that would help me with my example of build.lua :)

4 hours later…

5:24 AM

@PabloGonzálezL Hmm, that's odd: I'll see where the bug is

@PabloGonzálezL Hmm, that sounds like a bug too

@PabloGonzálezL That probably needs a new function: we've only had examples thus far that are typeset with one engine (either all-pdfTeX or all-LuaTeX)

2 hours later…

7:16 AM

@PabloGonzálezL I can point to the examples ...

@PabloGonzálezL The core of l3build is the testing: everything else is really secondary :) (We started with some scripts for testing, and they need file copying, and things grew)

@PabloGonzálezL Useful feedback; I'll try to pick off the issues

7:30 AM

@PabloGonzálezL it simply returns a table with some data which are either a bit confidential or used everywhere:

return
 {
   ["email"] = "XXX",
   ["github"] = "XXX",
   ["name"]  = "Ulrike Fischer"
 }

8:22 AM

> If an h wore pants how would it wear?

user image

9:08 AM

quack

ooh it brought another duck

9:45 AM

@UlrikeFischer, @PhelypeOleinik I'll send that babel-related fix to CTAN today

@PhelypeOleinik Also the xparse one

9:57 AM

@JosephWright keeping ctan occupied:-)

@DavidCarlisle ooh

@DavidCarlisle DDoS?

just noticed optex (opmac for luatex)

10:12 AM

I missed 234432 :-(. I needed only 20 points and then got a bounty of 500. And I missed 234567 because of it too :-(.

@UlrikeFischer oh no

@UlrikeFischer obvious solution is to give me a bounty of 500 to hit your target

@DavidCarlisle ooh

@DavidCarlisle well that is a good idea. I should look for a suitable question.

@UlrikeFischer any of mine would be suitable

10:35 AM

@DavidCarlisle done, now lets hope that no mean person accept an answer before I get the 20 points ;-).

@UlrikeFischer ooh

11:05 AM

Can someone remind whether replacing #1 in a macro with its argument value (assuming for simplicity that that argument value is just an ordinary string), counts as expansion, or not?

@FaheemMitha No, it's not expansion

@JosephWright ooh appendix 1 of my thesis :)

@JosephWright Ok, thank you. I looked for a description of the process, and it looks like it is more like simple text substitution.

Phelype Oleinik

@JosephWright Thanks! Sorry about the missing tests...

Per this very detailed discussion from egreg - tex.stackexchange.com/a/335386/3406

Can someone give me a reference for \next? I looked in TeX by Topic, but couldn't find where he discusses its usage.

Not finding it in tug.org/utilities/plain/cseq.html

Phelype Oleinik

11:18 AM

@FaheemMitha It's just a temporary macro. Not different from \x or \@tempa

Oh, never mind. In the example it was user defined, I think.

@PhelypeOleinik Ok, thank you.

11:38 AM

@FaheemMitha it takes place during expansion, but the actual replacement of #1 isn't in itself an expansion step

@PhelypeOleinik or \blub

I'm using BibTeX in the simplest possible manner with the iopart-num style. This style does not include DOIs in the output but it does include URLs.

Is there a simple way to make DOIs appear and remove URLs? (Or even better: only remove URLs when a DOI is present)

What would be the procedure? Would I have to edit the style (in which case I won't bother), or is there a simpler way?

@DavidCarlisle ooh blub

Phelype Oleinik

@DavidCarlisle Not \zz?

@PhelypeOleinik ooh

@PabloGonzálezL Ah, you are trying to put an extracted file into the CTAN upload: they don't like that, so l3build checks to avoid it. The only derived files we are supposed to upload are PDFs

@PabloGonzálezL It would be in a TDS zip

12:00 PM

@PabloGonzálezL I'm reading your .lua file to get to grips with your issue

@JosephWright I think that was main reason I didn't convert bidi to l3build but used the existing shell script from Vafa, README.txt is generated by docstrip and getting l3build to accept that was more trouble than it was worth as I had a working build script, and was trying not ti change the source layout too much

@PhelypeOleinik far too important a command name to waste on a temporary macro

@JosephWright As a certain point I wonder if it makes sense for a trusted repository to obtain permission to upload directly into CTAN to make updates automatic :)

@WillRobertson it used to be that way, they would mirror well known sources on to ctan

@Szabolcs it probably wouldn't be that much of a change bst is simple enough to edit once you get used to it.

@DavidCarlisle I'm doing that now, it's going okay so far

Is there a way to check in the format.xxx functions what sort of item it's going to be formatted in?

I'm looking at this: math.cmu.edu/~gautam/sj/blog/20171114-bibtex-doi/habbrv.bst and it has:

FUNCTION {format.url}
{   doi empty$ not
    url empty$
  or
    { "" }
    { "\urlprefix\url{" url * "}" * }
  if$
}

so it skips printing the URL is there is a DOI

what I'd like is to still print the URL in certain types of item, e.g. in @misc

Hmm...

I guess I could use a separate, different format function for that item

12:17 PM

@Szabolcs yes, this. (but would a @misc entry have a DOI anyway?)

@DavidCarlisle Well, it's a software entry, and the style didn't support @software ...

I made a copy of misc now and called it software, then modified that

12:44 PM

GUYS I AM GOING TO GERMANY

10

user image

4

@PauloCereda ^^^ two good news at the same time ;-)

@UlrikeFischer ooh

@PauloCereda oh! And to Czechia as well?

@yo' Give some minutes, I will poke you. I am still in shock...

@UlrikeFischer you are really going to award 500 to egreg's typesetting service????

12:55 PM

@DavidCarlisle LOL

@DavidCarlisle sure - it is a important improvement to the default typesetting quality and there are no zz-commands in it, so I can actually use it ;-)

2

@UlrikeFischer I suppose it is the best answer of egreg's on the site, even if he didn't write it.

@DavidCarlisle I understand.

There isn't a fully expandable version of \NewDocumentCommand?

6

A: Is there an `\edef` version of `\NewDocumentCommand` or `\newcommand` with optional arguments?

Not sure what this would be useful for. \documentclass{article} \usepackage{xparse} \ExplSyntaxOn \NewDocumentCommand{\bunchdefine}{m} { \clist_map_inline:nn { #1 } { \exp_args:Nc \NewDocumentCommand { ##1 } { } { ##1 } } } \ExplSyntaxOff \bunchdefine{one,two} \begin{document}...

@FaheemMitha anything that has a name like that presumably defines a command, so by definition can't work by expansion, and must do an assignment

@FaheemMitha oh you mean like edef, not that the command itself is expandable?

1:11 PM

Is a \def definied inside a \def is only locally visible? I think I read that the curly brackets do something like scoping.

@DavidCarlisle Right, I meant like \edef.

@FaheemMitha no, commands inside a def are not executed at all

@DavidCarlisle I don't understand what you mean, but I could provide an example.

Anyway, it's easy enough to experiment.

@FaheemMitha \def\foo{ anything \zzzz \def\qqq{jjj} jjjj} you can not ask about the scope of the inner def as it isn't executed, just as you don't get an error for \zzzz not being defined

@DavidCarlisle One can use \qqq inside \foo, though.

@FaheemMitha you could but that isn't executed either

1:15 PM

@DavidCarlisle Oh. I must be missing something. You mean the macro \qqq isn't defined at all?

I could have sworn I'd seen such constructs.

@FaheemMitha not unless you later expand \foo and then evaluate its replacement text.

@DavidCarlisle Just calling \foo doesn't qualify as expansion?

@FaheemMitha yes that would expand \foo

@DavidCarlisle Ok. More experimentation in the offing, I guess.

@FaheemMitha but if you expand foo it's just as if you had anything \zzzz \def\qqq{jjj} jjjj in the file so \zzzz (most likely) gives an error, and \qqq gets defined to be jjj but you can't then talk about the scope being inside the defintion of \foo as the fact that the outer macro was involved is not really preserved.

1:25 PM

@DavidCarlisle As usual, not entirely following. But I'll try an example to clarify things.

2:02 PM

@DavidCarlisle it is much better than the answer I got the bounty for which basically says "I don't know the script but input a b instead of b a" ;-)

@barbarabeeton quack /giggles

@PauloCereda -- Yippee!

2:20 PM

@barbarabeeton <3

@PauloCereda -- Going to give a talk? (Don't even try to give an entire talk in German. You'll only get the same criticism as I get from Frank: "Your accent is execrable.") Just about everyone understands English, and they're very polite.

@barbarabeeton I would love to! The real challenge is fill out the registration form. :) Will do soon.

@MarcelKrüger you broke the missingchars testfile ;-)

2:39 PM

@UlrikeFischer It's fixed again. But that test is quite interesting: I never would have thought that Libertinus is missing ➍

@MarcelKrüger I was joking - it is a nice coincidence that I used the capital ß there. tu-tl2e7 breaks too. Should I correct it or do you want to do it?

@UlrikeFischer Please correct tu-tl2e7.

3:10 PM

@MarcelKrüger I'm just running the test suite and got a failure on arabic-gr which looks odd, on travis this test passed. There are differences in kerns and numbers: 0634 +0:+0 is changed to0634 +1:+0.

3:20 PM

@UlrikeFischer Oh, that's probably an effect of 0c7945cb4b: I tried moving some of the font scaling to HarfBuzz to simplify our code. It seems like HarfBuzz(or Graphite) gives different results on different platforms. Even if the change isn't actually visible, I will try reverting the change then.

@MarcelKrüger perhaps let me try again first. I seemed to have an outdated format in the way.

@MarcelKrüger no, still fails.

4:26 PM

@JosephWright Marcel added some nice code to luaotfload. When a font has no capital ß SS is used instead:

user image

2

4:42 PM

@UlrikeFischer Nice! (And I like this particular Eszett glyph, too!)

@MarcelKrüger ooh I will meet you there too

Pablo González L

5:19 PM

@UlrikeFischer Thank you very much, it is very useful, it makes me have "featuritis" and ask for a variable "personaldata=file" in l3build:)

@JosephWright I understand, using pkgcheck gives me the warning (and I remember receiving an email from ctan for this reason), this is a good point to add in l3build documentation, which reminds me that the note_file variable is very useful at this point...will it be possible to implement something like l3build upload -N ctan.note?

@JosephWright Thanks for taking the time, the only thing left for me to do is to compile example.tex using pdflatex (or latex>dvips>ps2pdf), that is, I want to have control over how I compile example.pdf. I haven't been able to get the typeset_demo_tasks()` function (which I think would be the right one) to do this.

2 hours later…

7:00 PM

Dear all, is there a cheat sheet of when to use ~ to prevent line breaks? I frequently run into questions like Lemma 1 and~2 vs. Lemma~1 and~2.

Also "The answer is~3." vs "The answer is 3."

Claude~E.~Shannon vs Claude~E. Shannon vs Claude E.~Shannon

7:43 PM

@UlrikeFischer I was positively surprised first time I saw this happen in ConTeXt.

2 hours later…

9:27 PM

@Skillmonlikestopanswers.xyz you complain if people don't use the ° symbol but see what happens if they do tex.stackexchange.com/questions/528483/…

9:37 PM

@DavidCarlisle ;-). Which made me look as siunitx and I wonder if @JosephWright will have to review the degree uses:

\documentclass{report}
%\usepackage{textcomp}
\usepackage{siunitx}
\begin{document}
\ang{40} \textdegree
\end{document}

@AlanMunn -- This looks more like something in your line, not an amsmath problem: How can I write an aligned environment with multiple anchor types?. Can you help?

@AlanMunn -- False alarm, sorry. The OP of that question has misled me, at least, by filling the example with all text.

10:43 PM

@mickep it doesn't work for me in simple document. Do one need to enable it?

1 hour later…

11:51 PM

@UlrikeFischer Probably not: \degree and \degreeCelsius need to match, the latter has to be made up of two glyphs, one a literal C, and that's not the case for \textcelsius ...

@JosephWright but the behaviour changes if one load the (now obsolete) textcomp.

@UlrikeFischer Hmm, that's more tricky, I'll admit

@JosephWright the \pm symbol seems to depend on textcomp too.

@JosephWright I couldn't test on an older system yet, but I suspect that one got different output depending on textcomp then, and so the question is if you want to retain somehow the old behaviour or have always the "textcomp is loaded" behaviour.

« first day (3393 days earlier) ← previous day next day → last day (1539 days later) »

Transcript for

Feb '2014

$TeX - LaTeX$ 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