1:28 AM
Comments you find in source files:
> Then replace the [lmodern] italic typewriter font with the oblique shape instead; the former makes my skin crawl. (Will, Aug 2011)

5 hours later…
6:35 AM
Does Lua offer anything like a Make equivalent, or should I just use Make?

@FaheemMitha There's nothing dedicated that I know of: the entire point is it's light-weight

@FaheemMitha l3build ... (@JosephWright)

@DavidCarlisle Well yes

Does l3build do building based on file timestamps like Make?

@FaheemMitha No, quite deliberately not

6:45 AM
I'm wondering how to integrate calling a function depending on some file timestamps.
No doubt there is an obvious answer.
I was to update an SQLite DB depending on whether the source file (which happens) to be a LaTeX file, is newer than the current SQLite DB file.
Simple enough, but I'm not sure how to do it.
Also not really on topic on TeX SE. Perhaps I should ask on U&L.
The SQLite loading is happening in Lua code, hence the Lua connection.

@FaheemMitha You'd have to use lfs: it can look up file attributes

@JosephWright Is that the LuaFileSystem library?

@FaheemMitha Yes

@JosephWright Ok, thank you.

@FaheemMitha For l3build, we wanted to avoid the risk of not recreating files or whatever, so we don't mess with file dates

6:49 AM
@JosephWright Is l3build home-grown, then? By TeX developers, I mean.

@FaheemMitha Yes: by ... me ... mainly

@JosephWright I see.

@FaheemMitha It's ultimately based on the older scripts we had for building LaTeX2e, but it's grown a lot

Make uses file timestamps, doesn't it?

@FaheemMitha I guess: I've never looked at the detail

6:52 AM
I've always considered those reliable.

@FaheemMitha I think I'd go with MD5 sums if I was going that way

@JosephWright Hmm. Not sure how that would work in a build framework. Wouldn't I have to record the MD5sums somewhere?

@FaheemMitha Yes, of course: it depends on what one wants to achieve, I suppose

@FaheemMitha they are distinctly unreliable with a git based workflow

@FaheemMitha If you want something generic and a bit like Make, l3build's not really it: the focus is purely (La)TeX files

6:55 AM
@DavidCarlisle Because version control messes with file timestamps?
@JosephWright Ok.

@FaheemMitha not version control in general, but git doesn't really preserve file properties it's based purely on the hash of the contents.

@DavidCarlisle Oh
I'm not sure if Mercurial preserves file modification times.

@FaheemMitha we tried cons for a while with latex (perl implementation of a make like system) but it was really painful so for some years things forked using make on *nix and home grown bat files on windows, but they didn't do the same things and making a release got impossibly difficult so then "we" (@JosephWright) re-wrote the whole build and test logic in cross platform lua (as shipped by luatex) so it works anywhere texlive works

@DavidCarlisle Yes, build systems can be problematic.

@DavidCarlisle Oh god, cons

7:02 AM
Doesn't look like Mercurial does preserve file timestamps. There's a filestamp preservation extension, though.

@JosephWright was that before your time in the team, I can't remember exactly?

I've used scons. I think that's related.
Possibly a successor. Written in Python.

@FaheemMitha yes I think we tried that as well, abandoned both.

@DavidCarlisle It was still about when I started; I had to use it a couple of times in the first LaTeX2e releases I was involved with
@DavidCarlisle Set it up on my Mac, there's a StackOverflow question I posted to get that sorted

7:25 AM

7:52 AM
Yesterday, on mathtools and not eating spaces, I was referring to
\MaybeMHPrecedingSpacesOff
\renewcommand\aligned@a[1][c]{\start@aligned{#1}\m@ne}
\MHPrecedingSpacesOn
Where \Maybe... is actually \MHPrecedingSpacesOff, defined from mhsetup. Here is the relevant code
\long\def\MH_nospace_ifnextchar:Nnn #1#2#3{
\MH_let:NwN\reserved@d=~#1
\def\reserved@a{#2}
\def\reserved@b{#3}
\futurelet\@let@token\MH_nospace_nextchar:
}
\def\MH_nospace_nextchar:{
\MH_if_meaning:NN \@let@token\reserved@d
\MH_let:NwN \reserved@b\reserved@a
\MH_fi:
\reserved@b
}
\long\def\MH_nospace_testopt:nn #1#2{
\MH_nospace_ifnextchar:Nnn[
{#1}
{#1[{#2}]}
}
\def\MH_nospace_protected_testopt:n #1{
\MH_if_meaning:NN \protect\@typeset@protect
\expandafter\MH_nospace_testopt:nn
\MH_else:
\@x@protect#1
The part that interests me is the fact that the aligned env basically does \newenvironemnt{aligned}{...\aligned@a}{...} so why does that don't eat spaces up to option arg work here, when we know it does not on a normal macro. Is it because of the expansion rules instead?

@daleif Because TeX is searching after a brace group, so not ingoring spaces; after a control word, TeX is ignoring spaces and there is nothing to be done

8:11 AM
@daleif in \begin{foo} [zz] there is a space token before [ so the macro code can choose to ignore it and take zz as an argument or choose to see it and typeset [zz] but with macros, \foo [zz] is the same as \foo[zz] there is no space token in either case..

8:23 AM
Thanks for the explanation. So it basically means that because \aligned@a is called from within an environment it can see a space token (because it gets delivered like this by the expansion), but in plain macros it is impossible as the space token is never seen.

@daleif more than the space token not seen, the space is never tokenised. Note this only applies to the first argument if you have an optional argument in any other position like \foo{zz} [bar] then you again have the option to handle the space token in different ways.

1 hour later…
9:43 AM
Wow! Four minutes and a spam post has disappeared!

@egreg woo

10:27 AM
This is probably only marginally related, but latexmk is unhappy that my LaTeX document doesn't generate a PDF. But it's purpose is only to generate some text files, using filecontents.
Is there some canonical way to get it to stop complaining?
The man page isn't the greatest.

@FaheemMitha -pdf- ?
Hm then DVI mode is enabled. Scratch that.

@FaheemMitha can you provide an example for testing

@FaheemMitha latexmk is really not the tool for that; you can simply run latex

11:02 AM
@FaheemMitha the manual has a section about output types "ALLOWING FOR CHANGE OF OUTPUT FILE TYPE".

11:12 AM
@daleif A file with just a filecontents block (I'm using the package) and nothing else should work. I used latexmk -pdf %s.
@UlrikeFischer Ok, I'll look at that. Thank you.
@UlrikeFischer Don't get a match for man latexmk. Version 4.61.
Is there a different manual?

@FaheemMitha texdoc latexmk

@PauloCereda Ah. Thanks.
Over here that just brings up what appears to be a config file.

@UlrikeFischer could you give me a pointer regarding MiKTeX?

11:32 AM
@PauloCereda Sure.

@UlrikeFischer Could you find out the path for arara rules?

@PauloCereda how?

@UlrikeFischer I've no idea (feature for next release: print rule locations). :)
@UlrikeFischer just check if it's not AppData\Local\Programs\MiKTeX 2.9\scripts\arara\rules.

@FaheemMitha otherwise make a question on the site or email John Collins (the maintainer), John reads the site. What is the purpose of a LaTeX file that only generates other latex files (other than perhaps dtx files)

@PauloCereda hm. I'm not sure how to check if something is not in the path. The main rules are on my system all in D:\MiKTeX2.9\scripts\arara\rules. AppData\Local wouldn't be a normal installation place, but it is possible that someone put miktex there. But why do you need external info? Doesn't arara know where it looks for rules?

11:48 AM
@UlrikeFischer It does, I had a user who set the rule path to that location, that's why I was intrigued. But everything worked out in the end. :)

@PauloCereda does arara use kpse and all the trees?

@UlrikeFischer Not yet. :) arara uses itself to find the closest rule path.

@PauloCereda well an option "show the path for rules" would be useful. I always forget where I put some rules needed for the christmas stuff shared by my computers.

@UlrikeFischer ooh

12:18 PM
@PauloCereda While I was looking at it and before I could reach the flag button, the counter went from -4 to -7 and the post was removed!

12:44 PM
@yo' Spotted \everymath{\displaystyle} in a class for theses.
@yo' And, to be really sure to get bad typesetting, \let\prelim\lim\renewcommand{\lim}{\displaystyle\prelim}

1:20 PM
@barbarabeeton a surprise quack!

@daleif In this case, it just outputs CVS files. It's just a convenience thing.
That file also has comments.

1:50 PM
@UlrikeFischer: now I realized that most of the info I wanted to know is already in the header of the arara.log! It seems my past self is way less dumb than my present self. :)
2

2:33 PM
@egreg Haha

Most of here have ailihphilia. Hopefully nobody suffers from aibohphobia.

@daleif I'm not sure what to do. Ask a question on the site? Write to John Collins? What do you advise? Seems like a trivial question for the site - is it really suitable?
It doesn't look like latexmk has a public issue tracker.
And unlike most things, it's not on GitHub.
And CTAN doesn't have an issue tracker or repositories. Actually the CTAN page doesn't even mention the upstream software home page.

@FaheemMitha CTAN is simply a code clearing house, nothing more. It's up to individual package authors to supply any information about bug trackers or repos.

@AlanMunn An upstream page would be useful, at least. Debian includes that in package information, for example.

2:51 PM
@FaheemMitha I know, but it's up to authors to provide the info. You can when you submit. See the optional fields here: ctan.org/upload

@AlanMunn I see. Thank for the clarification.
Weird, it's defaulting to latexmk here.
Oh
> The upload has been started after a package page for latexmk has been visited.
That's very thoughtful.

@FaheemMitha E.g. @JosephWright has info for siunitx ctan.org/pkg/siunitx but @DavidCarlisle has none for his 88 packages... :) (Well I didn't check all of them. Maybe there's a bug tracker for indentfirst.)

@FaheemMitha But for a lot of people it's just 'email the address in the source'
@FaheemMitha CTAN can only add data as supplied: GDPR and all that

@AlanMunn @DavidCarlisle doesn't need bug trackers, he just needs feature trackers.
7

@JosephWright I wasn't suggesting CTAN add the web pages themselves. I wasn't aware that GDPR was relevant, though.

2:59 PM

Would anyone like to comment on whether the latexmk would actually be a reasonable question for the site, please?

@FaheemMitha I guess it's borderline. If it can't be done, it probably falls into a feature request, in which case emailing John would be a better route than asking a question.

@AlanMunn Hmm.

Gah! Why does someword hyphenate, but someword--- not hyphenate with lualatex? :(

3:10 PM
A best regards everyone into chat
I'm choking by the heat. Now it will be 33 degrees or 32. I hope best for yourselves in chat and wherever you are.

@Sebastiano Come back to me when you get to an Adelaide summer…

@FaheemMitha On the other hand, given that there are more than 12 pages of options, maybe some combination of them actually can do it, and asking a question would get such an answer.

@FaheemMitha If you feel it is worth looking into just write him an email, it is listed in the manual. I do not have a github page or a bug tracker for memoir either.

@Sebastiano, in fact we will get to 30 here tomorrow!

Hi @Sebastiano
@AlanMunn Ok, I'll ask a question then.

3:58 PM
@FaheemMitha Why didn't you give a name to your generated file?

4:15 PM
@PauloCereda Oh, I forgot. I'll do it now.

@AlanMunn there is (the latex project github), but no one ever reported anything of course

4:59 PM
@DavidCarlisle Would you like people to?

@DavidCarlisle there is also a dpc github where no one ever reporting anything either ;-(

@DavidCarlisle I wasn't actually making an existence claim of the repo, but an existence claim of a link to the repo in the CTAN listing.
@FaheemMitha Surely a DPC bug is like a DPC question.

@AlanMunn Mythic?

5:26 PM
@AlanMunn yes I know, I noticed while updating textcase the other day that there were no links and I added them for that one, but the others not yet.
@FaheemMitha there is not really room for a bug in indentfirst (the package code is only 4 tokens)

@DavidCarlisle hmmm

6:10 PM
@DavidPurton -- "someword---" wouldn't hyphenate even with plain tex. it follows the rule "don't hyphenate a word that already contains a hyphen." If you want it to hyphenate, insert \hspace{0pt} before the hyphen, probably followed by \nobreak so that the line won't break before the em-dash.

2 hours later…
7:44 PM
@Sebastiano We have 8 °C right now …

@HaraldHanche-Olsen luxury

@DavidCarlisle Indeed, I can chill my beer outside. No need for a refrigerator.
@DavidCarlisle I found a bug in a one line shell script once. And that one line was short. it was part of Solaris.
The script was /usr/bin/kill, and the one line was something like kill "\$@". (In addition to several comment lines containing a copyright notice.) The most important bug: It lacked an initial #!/bin/sh line.