@UlrikeFischer Fixed. The problem was that Travis doesn't have the font installed, so it had to use the font from supporttexmf. But the texmf.cnf enabling supporttexmf was only enabled for tests, not for the documentation part.

@Skillmon It's safer ...

@JosephWright although actually just using csname is safer than it was as the utf8 filenames support means that at least for characters they will detokenize themselves \ifincsname (@Skillmon)

@DavidCarlisle I was thinking more generally of arbitrary macros as kets

@JosephWright sure but if you use a key name with assignments in and it falls over that may be acceptable as a documentable feature, but saying keynames are letters that happen to take one byte in the utf-8 encoding of unicode is a bit less acceptable as a documentable interface.

colortbl just updated on ctan. @Skillmon was complaining about the colour choices, I just looked again at the manual, what's not to like about

@DavidCarlisle :)

@JosephWright git? :)

quack

@Diaa sorry, I was quite in a hurry yesterday. Let me catch up with the discussion...

@PauloCereda dinner

@Skillmon if you want to see painful colour choices, look at ctan.org they have dusted off their Halloween outfit and stuck some Chinese characters on it, it really is worse than the colortbl above.

@DavidCarlisle oh no

@DavidCarlisle oh, no. Another CTAN-is-ugly-day :(

@MarcelKrüger here an example that fails with the error Unknown multiscript table blabla. It looks as if the font feature the name is lowercased but not in the definition.

\documentclass{article}

\usepackage{fontspec}

\directlua{luaotfload.add_multiscript
 ("Blabla",{grek = "DejaVuSans:mode=harf;script=grek;color=0000FF;"})}

\begin{document}
\fontspec{Latin Modern Roman}[RawFeature={multiscript=Blabla}]
a π
\end{document}

@DavidCarlisle I think it's aligned with your food taste :)

@CarLaTeX bright and interesting.

@DavidCarlisle Exactly what I was thinking of...

@CarLaTeX let's see inappropriate juxtaposition of primary colours. Hmm whose avatar does that remind me of...

@DavidCarlisle I see only gorgeous avatars around :)

@Diaa I added an answer.

@DavidCarlisle what does the table try to explain? Those colors hurt my eyes lol

@manooooh it is demonstrating that it is possible to have background colour in table cells.

@Skillmon Sure: @DavidCarlisle's point about UTF-8 in pdfTeX is probably the most important

@Skillmon Sure

@JosephWright other points we should agree on: Communication channel and build tool-chain. Would you prefer communication via mail, or some kind of chat? Do you have any preferences on the build process?

@Skillmonlikestopanswers.xyz I’m flexible :) Give me until mid-week and I’ll get something drafted as a starting point

@JosephWright Do you intend to include plots of performance if we include a benchmark section? If so, I'd like to do those with Python's Matplotlib (if I were to create them). Or do you prefer other plotting tools (I also speak a bit of Octave if you prefer that)?

@JosephWright also: take your time, afaik, we have still plenty of time until the deadline to hand in new articles, and I should be busy the coming weeks as well :)

@Skillmonlikestopanswers.xyz I do my plots in TeX, oddly enough ;) But as you are sorting the data, it’s your call on how to create them

@Skillmonlikestopanswers.xyz I’ll work on things; I’ve got a structure in mind, once it makes a bit of sense I’ll send it to yju

@Skillmonlikestopanswers.xyz ooh mr. rabbit has a cool address

@PauloCereda I feel a bit bad exploiting the user name for advertising, tbh.

@Skillmonlikestopanswers.xyz quack <3

@JosephWright I like to do them in Python for the computation facilities and then export from Matplotlib via PGF to PDF, such that the plot looks just like the rest of the document.

@JosephWright only problem is, that this requires to know and have control over the final formatting of the document, after all the plot creation has to use the same settings. Maybe I'll try pgfplots or similar for this :)

@Skillmonlikestopanswers.xyz I just made a joint document and we avoided discussion by mail but used the github issue tracker instead (in a private repo). It helps to organize topics.

@Skillmonlikestopanswers.xyz I guess I should join ;)

@Skillmonlikestopanswers.xyz we can work on the detail later

@JosephWright seems to share developers with sile:-)

@JosephWright @DavidCarlisle Do either of you use l3build upload --file=foo.txt ? Should I expect to need to escape backslash in there (e.g., write \\rmfamily)? CTAN have complained about my uploads in the past and reasonably sure it's not my preprocessing that is to blame...

@WillRobertson not since I was testing it. You could set the debug flag so it posts to that http post echo to see what it will send.. (but if it does need \\ I think that's a bug)

@DavidCarlisle With --debug I get something like the following, where the backslash is indeed missing from \epigraphnoindent:

  "form": {
    "announcement": "Try to prevent breaks after epigraphs, and add epigraphnoindent to automatically suppress indentation after all epigraph commands.",

whereas:

cat epigraph-ctan.curlopt

form="announcement=Try to prevent breaks after epigraphs, and add \epigraphnoindent to automatically suppress indentation after all \epigraph commands."

@UlrikeFischer never used GitHub that way... Good idea, thanks.

@JosephWright I'd love to see you guys there :) But I'd understand anybody saying "oh, yet another Q&A, guess I'll pass". IMHO, SE and their current course of action make an alternative necessary.

@Skillmonlikestopanswers.xyz it worked quite well. emails can get messy quite fast, even more if people use different quoting styles (I ended with around 1000 github notifications for around 50 topics, I shudder considering to have the same number of real emails to handle). Also you can edit issues. So you can open an issue "Don't forget XYZ" and write "details later". And you have a project management and can assign people.

YAY WILL IS HERE

@PauloCereda Not for long :)

@WillRobertson -.-

@WillRobertson oh no

@UlrikeFischer Yes, issues are pretty good for structured information exchange. But I think a chat is faster for short termed information.

@Skillmonlikestopanswers.xyz yes, that is fine as additional communication. We used also skype calls.

@PauloCereda Do you know how to quote strings being sent to curl? A single backslash disappears, and two backslashes end up as two. Confusing.

@PauloCereda only to complain about l3build features :-) (@WillRobertson)

@WillRobertson perhaps %22 works?

@DavidCarlisle ooh

@DavidCarlisle thankfully my experimental branch is not working, so you could blame me. :)

@WillRobertson I had a terrible job getting newlines to survive at all, especially as it depended on which commandline is in use, that's why it goes via that file.

@DavidCarlisle and as I'm using windows ;-)

If  the  parameter  contains whitespace (or starts with : or =),
              the parameter must be  enclosed  within  quotes.  Within  double
              quotes,  the  following  escape sequences are available: \\, \",
              \t, \n, \r and \v. A backslash preceding any other letter is ig‐
              nored.

@WillRobertson so \\ ought to work, I think....

@DavidCarlisle someone suggested triple double quotes.

"""ooh"""

@DavidCarlisle people should omit any sort of description or documentation, nobody reads them.

@PauloCereda that is a man page not actual real typeset doc

@DavidCarlisle oh no

@UlrikeFischer so am I (not that you can tell)

@DavidCarlisle "boo"

@PauloCereda because every windows user uses a bash shell and man as the main help system.

@DavidCarlisle ooh

@barbarabeeton oh no twitter.com/notfunnyelle/status/1220926612905975808

@DavidCarlisle I thought you were in denial ;-)

@PauloCereda the fact that she wants to feat duck bread tells me she's a horrible person. Why does she want to poison those poor ducks?

@Skillmonlikestopanswers.xyz you shouldn't poison ducks as the poison will get into the food chain

@DavidCarlisle hence you shouldn't feed ducks bread. It's unhealthy for them and lowers water quality of the ponds.

@PauloCereda and @Skillmonlikestopanswers.xyz -- Yes, indeed, that is a horrible practice. A potential duck-exterminator. Don't follow her example.

@Skillmonlikestopanswers.xyz oh no

@barbarabeeton oh no

@DavidCarlisle oh no

@Skillmonlikestopanswers.xyz topanswers seems like latex.org or other sites that have been tried. They are not in themselves a bad thing but having one "go to" site for answers is a big win and having others dilutes the resources making answers at any one site.

@DavidCarlisle latex.org is a forum, not a Q&A. AFAIK, there is no other big English Q&A about TeX, only some smaller language specific one, like texwelt.de or technique.fr. I think one of the bigger problems with SE is that they don't have any competitors, so they can dictate the rules. If there is a competitor things on SE might turn to the better. I personally like the approach taken by topanswers and think it is a huge advantage that it is a non-profit, and I hope it will become a competitor.

@Skillmonlikestopanswers.xyz SE are not making any profit from any of the 'smaller' sites, so I think one has to be cautious

@Skillmonlikestopanswers.xyz it's a view but as I say I think having one site is actually a feature not a problem. similarly to github/gitlab/bitbucket/whatever, we can arbitrarily pick on github and use it as issue tracker with a reasonable chance people can comment, we can hang around here with a reasonable chance people can comment, if there are multiple sites with separate user accounts who does it benefit?

@Skillmonlikestopanswers.xyz at least with stackexchange it's clear who does run it. I looked around topanswers.xyz and found no statement at all about ownership or governance. At the github site you can see the individuals who have made commits, but that is about all.

@Skillmonlikestopanswers.xyz where do you feel that they dare to "dictate" a rule only because they don't have a competitor?

@UlrikeFischer perhaps not only because they don't have a competitor. But the recent developments don't point towards a reasonable direction, they ignore their user base and change things nobody wants changed except for them.

@Skillmonlikestopanswers.xyz well apart from time where they made the middle column too small I wasn't affected by some negative change. And I think you can't really know what the user base want - don't forget that on the meta site discussions you see mostly the people who have a problem and are against a change and not the people who are getting along well.

@UlrikeFischer yes, there most likely is a bias.

@Skillmonlikestopanswers.xyz I have watched some of the discussions on meta with somewhat detached interest but apart from one or two people out of a user base of millions who were directly impacted I don't think any of it has any affect on the users of the site.

@Skillmonlikestopanswers.xyz and all the issues are nothing to do with the stack exchange company really just inevitable social friction when managing user interactions on this scale. It is easy to start up a new site with half a dozen users and claim it runs more smoothly but if it has a decade of success and millions of users then the same issues just come back

@DavidCarlisle never claimed it'll work on a large scale, just expressed my hopes that it will. I don't have the definite answer :(

@JosephWright how come ctan has me down as maintainer....

\documentclass{article}

\usepackage{setspace}

\singlespacing

\begin{document}

\showthe\abovedisplayskip
xxx
\[\showthe\abovedisplayskip a=b\]
\end{document}

@Rmano still skiing?

@UlrikeFischer yep, tomorrow a bit more. Good snow but bad light... We hope tomorrow will be better.

@Rmano where are you?

Spanish Pyrenees, in Candandhu

@Rmano oh, looks quite nice.

Not too bad, nice destination for a weekend

@CarLaTeX when we came home I found a letter for the Bär with a shirt:

@UlrikeFischer ooohhh

@UlrikeFischer good day for Gladbach!