« first day (3246 days earlier)      last day (157 days later) » 

8:09 AM
Folks, I am looking for ways to include audio files in PDF file. I found a couple of old answers such as this: tex.stackexchange.com/questions/7502/… Is there any better/recommended way to do this?
@GermanShepherd Are you targetting Adobe Reader or do other viewers have to be considered?
I usually use the Document Viewer
@JosephWright yes, Adobe Reader as well
@GermanShepherd I meant there are lots of things that only work in Adobe Reader, at least reliably
@GermanShepherd Look at the media9 package
@JosephWright, yeah, media9 was what I was thinking of as well!
@JosephWright if I have an internal expl3 command __my_whatever: that is not expandable and is used only inside a \NewDocumentCommand, should I use \cs_new:Nn or \cs_new_protected:Nn?
8:25 AM
Consider this question: tex.stackexchange.com/q/509037/3929, I know where the error is, multlined used a blank line that it needs to back up, it does so via \vskip-\baselineskip \vskip-\normallineskip, but inside the table cell \baselineskip is 3pt and outside it is 15pt. Why?
@UlrikeFischer protected
@UlrikeFischer Everything that is non-expandable should be protected as a matter of course: there are very few places where it's an actual issue
@UlrikeFischer Bruno and I have made sure that (almost) everything in expl3 is either protected or x-type expandable: there is the business of \halign` support, but unless you are in that area it's really quite simple to make the call
@JosephWright thanks, I thought so but wanted to be sure.
@UlrikeFischer Cool: this was something I picked up from ConTeXt years ago
@UlrikeFischer Avoids any surprises if something gets used in a 'wrong' place
8:44 AM
@JosephWright well in my case this can't really happen - it is simply the auxiliary function from a user command. But I think it is quite good to explicitly consider if a function is expandable or not.
@UlrikeFischer Sure, but there is no cost to making things protected and it makes a clear line
Does the LaTeX project (or any of the other TeX projects) have standards about the format of log messages? This is something that often (usually?) seems fairly loose.
This is something I seem to find myself thinking quite often about. Possibly it's my latent OCD kicking in, and I'm making too much of it. But clear and modestly detailed log messages can occasionally be quite useful. As in, when you happen to need to look at them, they are useful.
For example a general convention is to have a first summary line (and in some cases, without a period for some reason). Followed by a blank line, followed by more detail, if necessary. But of course, how the detail is presented, varies.
@FaheemMitha Yes, to some extent: all LaTeX kernel messages are formatted the same way, and we recently standardised the expl3 ones in the same form
@FaheemMitha LaTeX kernel messages wrap with (package-name) as a line-continuation indicator
@JosephWright Ah. Is this documented somewhere public? Like some standards document?
@JosephWright I'm not sure I follow that. You mean you have (package-name) on all the lines of the log message?
9:40 AM
@FaheemMitha It's more a consequence of the fact that everything comes from \PackageError, \PackageWarning, \PackageInfo
@JosephWright You lost me there. But, regardless, don't feel obliged to clarify.
@FaheemMitha Almost all messages in LaTeX are generated by those mechanisms, and they provide a certain type of formatting
@JosephWright Log messages too? Because that's what I was asking about.
@FaheemMitha Yes: that would be \PackageInfo
@JosephWright Oh.
Are you saying that commit logs are machine generated? Or am I misunderstanding?
9:48 AM
@daleif why 3pt (I would have expected 0pt)
@FaheemMitha Huh? I was thinking of the .log generated by a LaTeX run. Where you thinking of something else?
@JosephWright Yes, version control commit logs.
I see I wasn't originally clear/explicit enough. Sorry about that.
@FaheemMitha git has some standards about that (it uses the first line in sumaries, so the idea is to keep that short and if needed add more on later lines) also github has rules about triggering behaviour like closing issues by putting keywords in the commit log.
@DavidCarlisle Yes, Mercurial does similar things too, I guess. Though Mercurial has no direct GitHub analog, so that doesn't apply.
But I really interested in more detailed policies/standards. Like if commit messages are organized wrt changes to files and/or macros/functions.
@FaheemMitha well there are still some public website project trackers that support mercurial although true most use git these days
9:53 AM
@FaheemMitha Not really: we are quite laid back about things
This is LuaTeX, Version 1.10.1 (TeX Live 2019)
 restricted system commands enabled.
LaTeX2e <2018-12-01>

luaotfload | load : FATAL ERROR
luaotfload | load :   × Failed to load module "luaotfload-multiscript.lua".
luaotfload | load :   × Error message:
luaotfload | load :     × "module 'luaotfload-multiscript.lua' not found:".
luaotfload | load :     × "no field package.preload['luaotfload-multiscript.lua']".
luaotfload | load :     × "[kpse lua searcher] file not found: 'luaotfload-multiscript.lua'".
@FaheemMitha Some projects have very restrictive commit rules, but the team don't
@UlrikeFischer ^^^^
@DavidCarlisle No really big ones left. Bitbucket is due to discontinue Mercurial support next year.
@FaheemMitha you may be the last person on the planet who hasn't switched from mercurial to git:-)
9:55 AM
I believe contingency plans are being made by Pierre-Yves David and the rest of the Mercurial team, and probably other people, but I don't know of any details.
@DavidCarlisle Happily, that is not the case.
Possibly the only person in this room. There don't seem to be any others here. Or if there are, they haven't said.
Some people are switching to sr.ht, but I'm not sure if it would take off in a big way. I think something that was more community supported would be nice.
sr.ht seems to be a one-person effort.
@JosephWright ooh I can remove stuff
@FaheemMitha there is a reason bitbucket etc switching to git, if the costs of maintaining a version control system that (say) less than 1% of users use is too high, then they will stop offering this option. It isn't just this room
@FaheemMitha A bit like BerliOS used to be?
@FaheemMitha What sort of thing were you thinking?
@JosephWright oh geez no
@DavidCarlisle The reason they're stopping is complicated. But in large part it's because they have been screwing Mercurial users over for quite a while. So people were unhappy. I think their response was to shut down support instead of trying to make it better.
I don't know how many people are still using Mercurial, but I think there are quite a lot. Though many of them are now doing private hosting. I hear that's quite common within companies.
10:00 AM
@DavidCarlisle also we have gti in the command line to make fun of us when mistyping git. What has Mercurial ever done for us? :)
@JosephWright I'm not familiar with BerliOS.
@FaheemMitha It vanished a few years ago
@FaheemMitha I used to host my code there
@JosephWright It's basically about how to best organize a commit message.
@JosephWright Oh. What happened to it?
@JosephWright And how was it?
One basic question about a commit log message is how much information/detail to include. And then how do you organize/present that information so that your future self or someone else will find it useful?
@HenriMenke ups.
@FaheemMitha It was a one- or perhaas two-person operation. Closed down, all the repos (SVN) had to move
@FaheemMitha It was OK: BitBucket was better when I moved there, but it was a marginal gain
@FaheemMitha Like I said, we do not go for the super-organised approach, but then we also are not restrictive about what is in each commit
10:16 AM
@HenriMenke Hm. I just updated luaotfload and the lua is there and found: Lua module: luaotfload-multiscript 2019-09-13 3.00 luaotfload submodule. c:/texlive/2019/texmf-dist/tex/luatex/luaotfload/luaotfload-multiscript.lua.
@FaheemMitha There are projects where you have to specify if the commit is an addition, a change, whatever: we don't do that
@JosephWright hmmm
@JosephWright aw
mktexlsr: /usr/local/texlive/2019/texmf-config/ls-R: no write permission, skipping...
mktexlsr: /usr/local/texlive/2019/texmf-dist/ls-R: no write permission, skipping...
mktexlsr: /usr/local/texlive/2019/texmf-var/ls-R: no write permission, skipping...
mktexlsr: /usr/local/texlive/texmf-local/ls-R: no write permission, skipping...
mktexlsr: Done.
@UlrikeFischer ^^^ Found the problem.
So it's on my side.
@FaheemMitha, @PauloCereda chris.beams.io/posts/git-commit is a reasonable position
@FaheemMitha @JosephWright git log --oneline from a public project that I could mention...
10:20 AM
@JosephWright We could also pipe random messages from whatthecommit.com :)
7b520ec (HEAD -> master, origin/master, origin/HEAD) updates
73784ec updates
f5cb083 updates
1c3ee7c updates
9ef5c10 updates
5712c4d updates
a7e39de updates
eaa83b1 updates
2f489a1 updates
2aac246 updates
2fc2fa3 updates
4e6a03e updates
aba6dd0 updates
91ca466 updates
68cbd88 updates
d461c0f updates
689e897 updates
df12c36 updates
cf95052 updates
745ff4f updates
2f8bf14 updates
a8e23ad updates
2846391 updates
98fe88e updates
9ed903e updates
cec7a30 updates
e1385cb updates
5937d73 updates
ef88260 updates
@DavidCarlisle OOH
rigorously enforced commit log style....
@DavidCarlisle and easy to parse
@FaheemMitha Although we don't have a formal set of rules, most of our commits are reasonably clear in terms of message. For example, we almost always reference issues when the commit closes one, as GitHub picks that up.
10:31 AM
@JosephWright Git on BitBucket?
@JosephWright I see. I think svn used to have quite strict rules. Maybe they still do. But it's been many years since I used it.
@JosephWright That's an interesting post. Thank you. I see he couldn't resist the lure of xkcd, though.
@DavidCarlisle Is that you, or someone else?
Here is a representative (but rather longer than average) log commit message.
changeset:   3703:f4a1a749f56e
parent:      3650:effc3f94d9b8
user:        Faheem Mitha <xxxxxxxxxxxxxxx>
date:        Tue Sep 17 22:40:09 2019 +0530
files:       business/smyt/abbrev.lua business/smyt/formletter.lua business/smyt/pdf.lua business/smyt/smyt.lua business/smyt/smyt2.sty business/smyt/
Split out parts of smyt.lua into separate files: abbrev.lua, formletter.lua, pdf.lua, smyt.lua, and util.lua.

Move code out of function `formletter` in formletter.lua and into
util.lua as a separate function called `loadyaml`.
Unlike Git, Mercurial has local cset revision (integer) numbers.
@DavidCarlisle well, seems more explanatory than my commit notes...
10:46 AM
@FaheemMitha No, Mercurial at that time
@FaheemMitha No, SVN itself enforces nothing
@JosephWright I mean the project, not the software.
@PauloCereda Are those actual log messages?
@FaheemMitha Oh, right, could well be
11:04 AM
@FaheemMitha they could be
@FaheemMitha svn itself has no rules about commit logs, individual projects may have
@FaheemMitha me
@DavidCarlisle I'm typing it out from within multline I was wondering as well, isn't amsmath adding \jot somewhere? I believe that \jot have a default value of 3pt
11:32 AM
@PauloCereda I think you replied to the wrong message.
Some of those commit messages are actually quite funny. I just hope they were fake.
I particularly like:
> I stopped caring 11 commits ago
11:53 AM
@JosephWright In view of this here github.com/MiKTeX/miktex/issues/370. Does l3build ctan do some line ending normalization? I never got a complaint but don't know if I only had chance or if it is by purpose.
@daleif sounds likely may look later...
@UlrikeFischer Yes: use zip features to normalise line ends
@UlrikeFischer That is why we zip in two parts: text files then add binary files
@JosephWright ah, good.
@UlrikeFischer Keeps CTAN happy
12:13 PM
@JosephWright and that is important ;-). I checked a number of candidates. Imho the l3sys looks all ok (I even used one yesterday ;-)), and \mode_leave_vertical too.
@UlrikeFischer Yes: my issue with \mode_leave_vertical: is whether to add to l3prg
@JosephWright hm. Yes difficult question. it doesn't really fit anywhere. Perhaps to l3box? At least that does typesetting too.
@DavidCarlisle weeeeell, multlined is not an amsmath construction, I don't even think it uses any of amsmath. I have no idea where it comes from. IT does not help much on the problem at hand that the value is zero either
@UlrikeFischer Good plan
@UlrikeFischer Which one did you use?
@JosephWright sys_if_shell_unrestricted: tex.stackexchange.com/a/508920/2388
12:22 PM
@UlrikeFischer So add the shell function?
@JosephWright ? you mean all the various shell-functions?
@UlrikeFischer Yes
@JosephWright they look all quite sensible to me. But the other canditates in this section too.
@UlrikeFischer I'm working through: expect more commits
12:40 PM
@JosephWright we need a linguist in the team ;-)
@UlrikeFischer me!
@DavidCarlisle ooh
@DavidCarlisle oh no
@PauloCereda you'll have to stick to pineapple toppings in future.
1:34 PM
@daleif -- a lot of mathtools doesn't have any relation to amsmath except for "reusing" cs names. The concepts are definitely related, but not the implementing code. In a "previous life" I suggested reimplementing name-related functions from mathtools directly in amsmath, but with the caveat that the internals would have to use amsmath methods. (Might be worth looking at breqn for clues.)
1:51 PM
@barbarabeeton there is a lot of stuff from mathtools that has a better home in amsmath, and for the most par tporting them is not that hard. The syntax Morten used is not that sofisticated compared to have expl3 can nowadays.
@daleif -- Exactly what I was trying to say. And since the "pedigree" of breqn is that the original attempt was by Mike Downes, and then turned over to Morten, that's why I suggested looking at that. (Now that amsmath is in the LaTeX team's hands, maybe something can actually happen.)
@barbarabeeton To most people breqn is probably a dead fish, unsolvable (perhaps with the aid of lua).
@daleif Probably should get the team to talk to you
Found the 3pt though, it is commong from \spread@equation from amsmath, it is used internally in multlined
@daleif breqn is workable, I think, if we took the ideas and re-worked from scratch
1:58 PM
@JosephWright about?
@daleif Er, things, I'll mail
@JosephWright I'm not going to touch it (breqn). I hardly do much LaTeX work these days.
@daleif Email sent
@daleif -- What breqn was intended to do is still highly desirable, but it's a very hard problem. My opinion is that it needs more direction from the author, who is the only one who really understands what the math means. I don't blame you for disowning it.
@JosephWright interesting my webmail thinks you sent me an event invidation.
2:01 PM
@daleif ER, that's very very odd
@barbarabeeton And Morten is not really active any more. I did not contribute anything to breqn
@barbarabeeton I think the ideas are workable, but like I say I'd be tempted to start from scratch. Probably one for some wider discussion
@JosephWright there are several things that seems best to be implemented on the basis of say xparse. Other things can be done in plain 2e and could be added to amsmath. One could even split mathtools (I knot it is a bad idea), in the 2e part and an xparse/expl3 part.
I'm signing off for now, later
@JosephWright -- Good approach. Start with looking at the goal, figure out what's currently undecidable by a simple-minded automaton, and contemplate how reasonable hints by the author could help. You can probably find lots of good bad examples in the tex.sx archive.
@barbarabeeton I'll perhaps raise this with the rest of the team
2 hours later…
3:57 PM
user image
@PauloCereda @barbarabeeton ^^
@yo' WOW
@yo' Simply amazing, pal!
1 hour later…
5:03 PM
@yo' -- How's your pedal practice coming along?
5:51 PM
@barbarabeeton My new teacher changed about everything I self-taught about pedal playing. However, he also changed the fingering principles and stuff. (Note that we're rehearsing J.S. Bach/J.L. Krebs, which still follows baroque principles.)
@yo' -- Is this your first formal instruction?
@barbarabeeton Basically yes (considering the organ). I had couple rehearsals as a part of 3-day organ workshops.
@yo' -- Well then ... I hope you didn't teach yourself too many bad habits that a teacher will want to change.
@barbarabeeton I don't think so. Of course I know the romantic (piano) fingering, which is quite different from the baroque one. For the pedals, the biggest thing is that I like to play barefoot, which is inappropriate (and also wrong for some true reasons)
@PauloCereda come try it :)
6:18 PM
@yo' -- At least you didn't try to wear boots. One organist I've watched perform wore what looked like patent leather (ballroom) dancing shoes, with a strip of rhinestones between the (very thin) heel and the upper. Something nice and soft and flexible. Barefoot is understandable, but be careful of splinters.
@barbarabeeton I always wore socks. However, from now on it's shoes always. (Which reminds me, I have to buy nice shoes for home.)
2 hours later…
8:43 PM
Could someone with a MiKTeX installation try this, please: tex.stackexchange.com/q/509098?
@PhelypeOleinik I will look.
@UlrikeFischer Thanks :-)
@UlrikeFischer miktex thinking \relax is a filepath again?
@DavidCarlisle I just thought the same and was trying to figure out a way how to test this. I could try the \csname trick again.
@DavidCarlisle yes, looks like the same, it complains about csname. I will add a remark to the issue.
Am I confused, or does wrapping (for example os.execute) in an assert hide (or does not propagate) the error in Lua? I think I've been seeing this for a while.
8:54 PM
@FaheemMitha Huh?
@JosephWright Assuming there is an error, of course. What part of that is unclear?
@DavidCarlisle hm. TeX live doesn't like 5\csname either, and thinking about it, mf is not tex, so it shouldn'T ;-(.
@FaheemMitha An assertion is just meant to fail if it's not true, and that's what it does
@UlrikeFischer put some metafont on a file called relax in the current folder and see if it gets loaded...
@JosephWright Right. I still want to see the underlying error, though.
Because that has the real information.
8:55 PM
@FaheemMitha You do:
'wef' is not recognized as an internal or external command,
operable program or batch file.
test.lua:7: assertion failed!
assert(os.execute("wef") == 0)
@JosephWright Doesn't seem to be happening here.
I'll take a closer look.
@JosephWright just catching up on mail, invites for mondays and Fridays?
@DavidCarlisle I canned the Monday one, no?
@JosephWright didn't see that (maybe lost in 10001 checkin messages, OK Fridays then I take it:-)
@DavidCarlisle Been busy working on some tidying: sorry we ended up with quite a lot of commits
9:02 PM
@JosephWright Ok, here is an example I'm trying to debug right now. With the assert, I get:
[4])./tables.lua:15: assertion failed!
stack traceback:
        [C]: in function 'assert'
        ./tables.lua:15: in function 'loaddb'
        ./tables.lua:71: in function 'tables.PrintDocTable'
        ./tables.lua:108: in function 'PrintDocTable'
        [\directlua]:1: in main chunk.
\PrintDocTable code ..., "\luaescapestring {#5}")}

@JosephWright compared to how many commit and or ticket messages I get at work, it hardly registers:-)
4])./tables.lua:32: attempt to index a nil value (local 'pipe')
stack traceback:
        ./tables.lua:32: in function 'loaddb'
        ./tables.lua:72: in function 'tables.PrintDocTable'
        ./tables.lua:109: in function 'PrintDocTable'
        [\directlua]:1: in main chunk.
\PrintDocTable code ..., "\luaescapestring {#5}")}

The relevant message here is, of course:
> attempt to index a nil value (local 'pipe')
@DavidCarlisle I'm not sure if it a filepath problem. If I put the code into a file and then try to load it, I get an emergency stop too:
This is METAFONT, Version 2.7182818 (MiKTeX 2.9.7140 64-bit)
! Emergency stop.
<*> blub.mf
Which doesn't appear in the former message.
@FaheemMitha That's a Lua error, though, not one from a called program
9:06 PM
@FaheemMitha why the assert, perhaps you want (x)pcall to trap the error? Hard to tell what you are doing though just from fragments of error output
@JosephWright I take it back. It looks like it's hitting a different Lua error further along, because the assert isn't stopping it.
@DavidCarlisle Possibly. My knowledge of Lua continues to be negligible. Looking up pcall and xpcall (again).
@JosephWright Thank you.
@DavidCarlisle I wasn't expecting anyone to diagnose anything. I just got the impression that wrapping in an assert was losing information, but I may be mistaken.
@FaheemMitha Assert is a pretty blunt tool: it's meant for avoiding out-and-out disaster
9:21 PM
@JosephWright I understand. I want something to make the program grind to a halt if there is an error. But I suppose there might be other options.

« first day (3246 days earlier)      last day (157 days later) »