1:03 AM
@samcarter Is there a ubuntu or windows version of this presentation app?

6 hours later…
6:40 AM
216

Most of the time when presenting a talk people connect their laptop to a video projector. So they effectively have an extra screen at their disposal. Because of that many Software options for presentations use the extra screen of the laptop to display notes, time left for the talk, a preview of t...

2 hours later…
8:33 AM
@JosephWright This is great work! Thank you very much. Now I have to adjust the GitHub Actions workflow files to use l3build. Somewhere I also have a branch with a flattened doc tree. Would that help with the CTAN target?

@HenriMenke I've got to release the update yet with tdsdirs support - I take it you are happy with the feature
@HenriMenke A flattened doc tree would avoid that awkward question of the repeated main file (I suspect CTAN will not be happy), and it also means that only the sources have to be 'always available', so yes I'd favour that if at all possible
@HenriMenke With tests moved to a proper setup, the entire doc tree can be simplified a lot I think
@HenriMenke See latest commit to show how to do dvisvgm testing, inputting a file that sets stuff up but keeping the same .lvt files as other workflows. (That's a_bit_ tricky and I might try to make it smoother to save .tlg files.)
@HenriMenke Would you like me to flatten the doc tree in my branch?

9:16 AM
@Rmano ooh squirrels use Thunderbird too :)

5
^^^ apparently @JosephWright won :)

@samcarter ooh now he can work on siunitx

@PauloCereda yes... and the (finally) says that bugzilla.mozilla.org/show_bug.cgi?id=377621 is fixed, but my last update didn't have it, I think :-(

@PauloCereda siunitx tikz

@PauloCereda Well, I waited 15 years, so waiting from 78 (the one I have) to 96 (the branch it seems has received the fix) will be bearable...

9:32 AM
@PauloCereda party! :)

@Rmano ooh
@samcarter yay
@DavidCarlisle ooh

@samcarter do you know what is the fasted method to get a pdf without background from a svg?

@UlrikeFischer only one page or multiple?
@UlrikeFischer if only one then inkscape and manually delete the background

@samcarter one page.

(and in the next step: complain to these annoying documentclasses which automatically insert a solid background colour -- beamer, I'm looking at you!)
@UlrikeFischer If you want to sent me the file, I can try if it works

9:44 AM
@samcarter mail just sent ;-)
@samcarter But no hurry, I have to go now for a few hours.

@samcarter I blame @samcarter

@DavidCarlisle Oh, can I do that as well? :)

10:18 AM
@HenriMenke, @MarcelKrüger l3build with tdsdir off to CTAN

10:33 AM
I work mainly on TeX LaTeX StackExchange. On a desktop machine running Debian Buster with Chromium 83.0.4103.116. Since yesterday nothing works anymore: When clicking on "Ask Question" or "Answer" or "Add Another Answer" nothing happens. Clicking on "Add comment" scrolls to the top of the page. When clicking on (+1)Upvote/(-1)Downvote nothing happens at all. Clicking on navigation elements no longer leads to the usual sub-pages but to stackexchange.com. What is wrong?

3

For years I've been browsing the site using the SeaMonkey Gecko-engine-based browser. It has the following user agent: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:68.0) Gecko/20100101 Firefox/68.0 SeaMonkey/2.53.10 Suddenly this afternoon, interacting with most of the site of the site seems broken: ...

(this does not only affect seamonkey but many other browsers as well)

How to find out about browsers that can be used?

10:59 AM
@HenriMenke I'm working out the best 'flat' structure

@JosephWright will be flat?

@PauloCereda Quick, hide the Earth or he'll make it flat was well!

@samcarter Thankfully Joseph is a chemist. If he were a physicist, I'd be panicking now. :)

11:15 AM
@PauloCereda He can blow it up, can't he?

@samcarter oh no
/panics

11:39 AM
@HenriMenke I've got things working without tdsdirs for the doc target, so no CTAN worries for the extra file. Also means there's no ambiguity in which files are used for typesetting. I think I'm about done: CTAN have the l3build update so it should be in TL by the weekend
texlua \$(kpsewhich l3build.lua) ctan  829.08s user 62.11s system 90% cpu 16:27.52 total

12:17 PM
@samcarter They must be kidding! Chrome/Chromium 95? Iirc Chromium 95 is from October this year, i.e. about 2 months old. Chromium 83.0.4103.116 is from Monday, June 22, 2020. I can't update every day. So it seems I am kicked out now and won't be here any more for a while.

@UlrichDiez yeah, the list is very narrow. Not even long time stable versions. Given the audience they cater to, I would assume a bit broader spectrum than the absolute mainstream browsers...

12:43 PM
@UlrichDiez What is really annoying is that core functionality is broken even! I could understand when you had maybe some layout shifts or what when using a non-mainstream (or older) browser, but core functions should actually be working even when you have JavaScript or CSS disabled. I would actually consider this a basic law of UI/UX design, since otherwise you'd also get in trouble with overall accessibility of the site very quickly.
2

1:19 PM
@JosephWright I've rebased my own flattened-doc branch: github.com/hmenke/pgf/tree/flattened-doc

@HenriMenke I just flatten the output using l3build ;)
@HenriMenke Your call, of course: I'm just trying to give examples
@HenriMenke Do you want me to rebase onto that and suggest an l3build setup?

I like this user, good slogan :) stackoverflow.com/users/14671221/montresor
3

1:39 PM
@samcarter ooh

1:52 PM
@JosephWright I thought the flattened doc tree would make your life easier in l3build.
@JosephWright Eventually I want to move to a flattened doc tree anyway.

@HenriMenke We have to cope with many things :)
@HenriMenke I would go that way - it's much more sensible in the round
@HenriMenke Do you want me to adjust the l3build suggestion or will you post-setup?

Where does this new trend come from not to put \end{document} in your example code when posting a MWE on this site? I just don't get it ...

@JosephWright I'll merge the flattened-doc branch into master.
@JosephWright Right now this still requires copying doc/generic/pgf/pgfmanual-<engine>.cfg to doc/generic/pgf/pgfmanual.cfg.

@JasperHabicht imho many don't really try there example but only build it from copy&paste snippets and then it is easy to miss the \end{document} (or the \begin{document}).

2:18 PM
@JasperHabicht Check if it is really not there or is just hidden from view because it is in the same line as the code fence?

@samcarter Thanks for this hint! Well, in this case it is really missing. It is just the, like, fifth question in row where it is missing. I understand that it is easy to forget a line of code while condensing and copying and pasting the code to make a MWE ... I guess I just need to cope with it =)

@JasperHabicht Write it onto your Xmas wish list. Maybe you'll get a whole box of \begin{document} which you can then insert over the next year as needed :)

@samcarter This would be so great! Some \documentclass{article}s would also be highly appreciated!

@JasperHabicht :D

2:35 PM
@UlrikeFischer microtype didn't define \z I see

@DavidCarlisle ?

@UlrikeFischer texlive list undefined command \z

@DavidCarlisle oh ;-)
@DavidCarlisle what did it do with the catcodes??

@UlrikeFischer not sure microtype source is a bit opaque, I suppose I could look at dtx instead of sty but...
\documentclass{article}

\usepackage{microtype}
\title{A test}
\author{}
\date{}
\begin{document}

\begin{abstract}
x
\end{abstract}
\section*{Introduction}
x
\end{document}

@DavidCarlisle can you reproduce the problem? It didn't error for me (but I can have other versions)

2:38 PM
yes

@DavidCarlisle I have a local version interfering.

@UlrikeFischer \expandafter\scantokens\expandafter hmmm means putting \makeatletter before teh abstract works again

yes
oh

@gusbrs ah thanks I thought I had seen something similar:-)
@gusbrs Do you know what's it doing there? \scantokens looks to be a dangerous thing to play with at that point and adding \makeatletter locally (the fix in github) is just fixing one symptom I suspect.

2:52 PM
@DavidCarlisle Oh, I'm afraid I don't, I was just aware of the problem because I had also been affected. The problem was identified and fixed in dev at another thread: github.com/schlcht/microtype/issues/3. Perhaps you can chime in there.

@gusbrs yes I think I will, I gave in and looked at the doc,
% If we have a group, we need to rescan in order to allow, \eg,
% |\verb|, |verbatim| or |lstlistings| material: before reading
% the argument, make the backslash a letter to prevent spaces
% from being inserted after commands, and have line breaks preserved.
% The \nonetex\ version will fail in lots of cases.
I suspect that's a bad change

@DavidCarlisle If that's the case, that would be welcome. For microtype I'm just an user, I'm not acquainted with the code and the underpinnings, and I suspect they'd be out of my league if I tried. :)
I just know there's a new feature in the new version, which adds some patches for \item, toc and a couple of things more, which is acknowledged to be a work-in-progress, but has been taking some bites here and there.
Section 9 of the docs, "Being pedantic about protrusion", if you'd like the shortcut.

@gusbrs I left a comment in the issue (not very helpful one)

@DavidCarlisle Thanks. And, since you are the master of \z, I'm sure it is insightful. :)

@gusbrs \def\z @{\z@} would have solved all the issues

3:08 PM
@DavidCarlisle Let's add it to the kernel then.

@PauloCereda ohh :)

4:19 PM
@PauloCereda how to grab dinner without damaging the pool

@DavidCarlisle oh no

4:49 PM
Unsurprisingly, someone came up with the idea already.
I agree! I think that you could theoretically generate this information from parsing the log file with \tracingall. Anyone game? :) — Will Robertson Jan 20 '11 at 11:36

@user202729 which idea?

> I suspect that the correct approach would be to modify the engine itself, but I still don't know how the expansion would be displayed. tex.stackexchange.com/questions/61010/…
Something better than tracingall. A real debugger. (old questions are so hard to find.)

@user202729 ah that, put \pausing=1 and single step through gives you the stepping behaviour and tracingall usually gives you more information than you need, so between the two can't say I have ever really needed more

Okay, \pausing is new to me
Single step through what? The file?

@user202729 through each line, I don't use it much but it's there
@user202729 try pdflatex '\pausing=1 \input small2e'

5:42 PM
@samcarter -- A podcast you might enjoy, about the injection of satellite swarms into orbit: economist.com/podcasts/2021/12/07/…

@barbarabeeton Thanks! Perfect timing so I have something to listen to on the drive home :)

6:01 PM
@HenriMenke That's only for a change-of-engine, right? So from an l3build POV it's a non-issue (although I could cover it from one of the hooks)

@JosephWright here we go again :)

6:17 PM

2 hours later…
7:48 PM
@JosephWright I'm just trying l3build locally. There are some issues.
The dvisvgm tests fail.
There also seems to be no way to skip the tests when packing with l3build ctan.
Also I've already run l3build doc. Why is l3build ctan building the manual again? The PGF manual actually takes quite a long time to typeset, so if it could reuse an already existing artefact that would be a huge time saver.

@HenriMenke Nope
@HenriMenke You got the latest one?

@JosephWright Latest on TL, not latest on Git.

@HenriMenke This is all by-design: the idea is that we always start from a clean slate to avoid, as far as possible, creating a release that doesn't pass the tests/releasing documentation that's not up-to-date
@HenriMenke Ah, won't work then: you'll need the stuff I added this morning (it's been installed by CTAN, so should hit TL tomorrow morning)
@HenriMenke The problem is l3build can't tell if the PDF is up-to-date other than by building it - I can add some more switches to override this, of course
@HenriMenke I only released the tdsdirs stuff today, plus an extension to specialformats to allow the dvisvgm tests to work nicely

@JosephWright You could check the mtime of the PDF against the mtime of all the docfiles.
I see. Then I'll try again tomorrow.

@HenriMenke You could, but that wouldn't cover e.g. a change in the system TeX Live tree affecting the build

7:54 PM
@JosephWright I forgot to ask, can one now set another environment variable (different to texmfhome I mean) to specifiy the install texmf?

@JosephWright I think that can be assumed to be immutable. Comparing mtimes is essentially how make works.

@HenriMenke Yes: we really wanted to avoid how make works :)

But this also doesn't cover, say /usr/include/stdio.h
@JosephWright Why? What is the problem with make?

@HenriMenke Exactly: for the LaTeX kernel (which is the key product for l3build), we want to be sure of exactly everything
@HenriMenke It tries to be 'clever' rather than just rebuilding everything

@JosephWright make -B?

7:57 PM
@HenriMenke I'm used to building the kernel plus expl3: that's over an hour on my laptop, so pgf feels pretty quick
@HenriMenke Pass: I'm not a make user
@HenriMenke I meant that it had 'cleverness' built-in, whereas for l3build the starting point was 'write down exactly what you want to happen, have that happen every signle time you run the tool'

@JosephWright make -B (a.k.a. make --always-make) rebuilds everything unconditionally gnu.org/software/make/manual/html_node/…

@HenriMenke Different approach, that's all, really
@HenriMenke I guess one thing that shows at the moment is that there are only a small number of tests and one big doc, so the balance feels like it's out - adding more tests likely makes it feel more balanced (but of course long)

@JosephWright Hm, I wanted to have checks and manual + ctan in different GitHub workflows to run them in parallel and cut down on the total CI time.

@HenriMenke We do that :)
@HenriMenke For LaTeX, we do exactly that for arbitrary checkins, and only run the ctan target for tags

@JosephWright How? I thought I can't do ctan without check?

8:00 PM
@HenriMenke You can't, but you can do check or doc on their own
@HenriMenke I have been wondering if we might want to add an option for that: the original design was based around building locally, whereas on a CI like GitHub Actions it's a bit 'safer' to skip the tests

@JosephWright I don't actually want to skip the tests but running them in another workflow means that the ctan workflow is not aware of them.

@HenriMenke Sure, all I mean is some --no-check or similar option

8:22 PM

8:40 PM
@HenriMenke Perhaps mail the team, or post an issue?
@HenriMenke Is a check/doc split going to work for you?

9:42 PM
I'm not making this up: "Even if free, all non-hardware (software, applications, eBooks, web access, cloud services/storage, etc.) MUST be requested for purchase via the IT Readiness form and go through the requisition process."

@AlanMunn I can believe it

@JosephWright Well that may be the 'official' line, but I suspect the real plan is to make buying computer stuff so outrageously complicated that we'll just pay for it ourselves out of pocket.

@AlanMunn Ah, right
@AlanMunn We have that: our departmental secretary is using an Alienware system for her job ;)
@AlanMunn I use my Macbook Pro as that way I can do what I like: there is some serious lack of joined-up thinking

@JosephWright But do they seriously think I'm going to fill in a multi-page form full of questions I have no way to answer before I update R or TeX Live. And how would they know? Maybe in the Windows world this stuff is easier to control externally, I have no idea.

@AlanMunn Our managed systems at work have some software installed that prevents any non-authorised binaries from running; I hear oddly that our computing department is exempt
@AlanMunn I understand your cynicism, but I suspect the more realistic reason is that it's a policy written starting from a corporate mindset, imagining a very small number of programs needed and with a few highly-defined configurations only
@AlanMunn These policies tend to be written starting from what typical administrators need: Office + Outlook + (probably) Edge + some finance software + some business information package

10:00 PM
@JosephWright Yes I think this is definitely the case. But in the research environment it is simply unworkable.

@AlanMunn Yes: whilst we are not quite in the same place, things have moved much the same way
@AlanMunn I take it you are not locked out of things like email from personal PCs (as one might be in a corporate, where you can't even log in using webmail)

@JosephWright No, and I think there would be major revolt if that happened. And it's very hard for the university to control student computers. A friend of ours who's at Penn State (which is very similar to MSU) has something like that and so most faculty simply buy their own computers to avoid having to deal with a university computer that they can't update themselves.
@JosephWright So my cynicism is somewhat grounded in experience.

@AlanMunn Our place did at one point talk about not providing (undergrad) students with emails, so it would be staff-only on Office365 and students on GMail

@JosephWright We have Office365 for all. But I still don't use Outlook. :)

@AlanMunn I suspect we'd have the same issue; it works for corporates where IT provide a new PC every 3-ish years, look after all the IT hardware, etc.
@AlanMunn Same here
@AlanMunn Formally, Outlook on the mac is just webmail anyway ;)

10:07 PM
@JosephWright I guess that's right.

@AlanMunn Unless they changed it: behind the scenes, it uses WebDAV, whereas on Windows it uses the proprietary protocol
@AlanMunn Don't know about you, but for us getting an office PC is 'interesting', and getting lab/student ones is a case of finding 'recycled' systems, so ... less controlled

@JosephWright Because I use a Mac and I hate transferring everything to a new Mac, I get a new computer as infrequently as possible, although we're allowed a new one every 3 years. And since I haven't had one for quite a while I really don't know if the new regime is locking down computers more than before.