@DavidCarlisle Found on twitter ^^^

twitter.com/simongerman600/status/1307448138798686214

@CarLaTeX ooh

@CarLaTeX the English one is clearly wrong: apart from myself no one understands other languages so will not be irritated, just confused.

@DavidCarlisle Falkland

@PauloCereda breakfast

@DavidCarlisle oh no

@PauloCereda I win

@DavidCarlisle me linux :)

@PauloCereda AIX

@DavidCarlisle I suggested adding "I like pineapple pizza" for Italy

@DavidCarlisle really? Have you used it?

@PauloCereda I had an aix box on my desk first few years I was at nag

@DavidCarlisle how nice, I always wanted to try

It was IBM's, wasnt it?

@PauloCereda yes

@DavidCarlisle In my old lab, there was a Sun Sparc, it was fun. :)

@PauloCereda oh I used suns for many years when I was at Manchester

@DavidCarlisle ooh

@DavidCarlisle, @UlrikeFischer When do we merge l3color into l3kernel?

@PauloCereda from 87 to 95 or thereabouts tex.stackexchange.com/questions/109758/…

@JosephWright after the october 2e release?

@DavidCarlisle I guess: I'm going to sort out the final xparse stuff just before that, then we will be in a position to go for the colour updates safely

@DavidCarlisle ooh cool

@DavidCarlisle Just looked at the stats: Wednesday was busy

@JosephWright yes it seems to be a steady flow these days, I think that was the busiest day so far

@DavidCarlisle and @JosephWright would you say that \cs_new:Npn \my_TF_true:w \fi: \use_ii:nn #1 #2 { \fi: #1 } is less robust than \cs_new:Npn \my_TF_true:w \fi: \use_ii:nn { \fi: \use_i:nn } (and could you think of a case where this is actually making a difference)?

@Skillmonlikestopanswers.xyz more joseph than me these days I can't think of a case where it will make a real difference so I can't say which is more robust but there are bound to be edge cases I haven't thought about:-)

@DavidCarlisle perhaps I should ask Bruno, he's really quick with finding edge cases.

@DavidCarlisle Well, the second version is as robust as \exp_after:wN \use_ii:nnn \fi:, so if in doubt I'd choose that one, but if the first version is equally robust I'd choose that one as it's faster.

It is this list: n*(n+3)/2 = 0, 2, 5, 9, 14, 20, 27, 35,...

@PauloCereda I had an account on an AIX cluster for a couple of years. But I don't recall many differences. Although it was super fast for the time. And all the cluster machine names were characters from Shakespeare. :)

It is already hardcore difficult to give a row a certain style in pgfplotstable:

\pgfplotstableset{
highlightrow/.style={
  postproc cell content/.append code={
  \count0=\pgfplotstablerow
   \advance\count0 by0
   \ifnum\count0=#1
          % \pgfkeysalso{@cell content/.add={$\bf}{$}}
              \pgfkeysalso{@cell content=\textbf{##1}}%
    \fi},     },
}

@cis @UlrikeFischer answered already in a comment. Also highlightrow/.list/.expanded=\List works (It's in the line above .list in the TikZ manual). Nice, I didn't know you can nest operators.

@cis Well, beauty is in the eye of the beholder, but I find this .list/.expanded quite elegant (and it's pgfkeys, it's not pgfplotstable).

@Rmano I created a pure TikZ example:

@Rmano This is not elegant and not not-elegant, this means to know the correct command. :) Ok, more then I knew. But I want to work out an automatisation, not a single-case thing which is fine at one special MWE - I always think globally, not locally.

--------> tex.stackexchange.com/a/563522/46023

@cis pgfmath/pgfkeys are normally not expandable. Why should I fight with them, if there is a two line expl3 variant?

Yay, I have another TeX-strangeness :)

In the following, the two lines containing \TYPE behave differently:

\cs_set_nopar:Npn \tmp {~}
\TEST{token~if~space}{
  \TYPE { \token_if_space:NTF ~ {TRUE}{FALSE} }
  \TYPE { \exp_after:wN \token_if_space:NTF \tmp {TRUE}{FALSE} }
}

At least in m3token002.lvt I get an unclosed \if from the second line but not from the first (with some patches to \prg_new_conditional applied)

@DavidCarlisle I found an edge case in which it makes a difference: If there is no argument for \my_TF_true:w to grab its replacement will not be inserted and \fi: not executed, leaving an unbalanced \if: there. So the first would be less robust in something like the following { \if:w a a \my_TF_true:w \fi: \use_ii:nn { one argument } }. Admittedly that already contains an error but the effects of that error are more severe with the first variant.

@cis most programming languages (I use) index starting from 0.

@Skillmonlikestopanswers.xyz yes I was going to comment earlier that error behaviour would be different although I didn't trace the details

@Skillmonlikestopanswers.xyz well there is no space token to test in the first case

@DavidCarlisle Well, I did. What really seems odd is, however, that in the code above (the one from m3token002.lvt I get that behaviour only from the second line, not from the first (while if I use this in a standalone file, both fail).

@DavidCarlisle well, there is no space token to test in the second case either, it is ignored by TeX-argument grabbing logic.

@Skillmonlikestopanswers.xyz t\tmp expands to a space token though so it's different, you don't get a space token from ~

@DavidCarlisle Yes, I know, but that space token will be gobbled when \token_if_space:NTF grabs its first argument (since spaces are ignored that argument will be TRUE).

@Skillmonlikestopanswers.xyz yes I guess you have looked what \token_if_space:NTF is doing, and it grabs the N as an argument:-) (I haven't looked at the code at all so just looking at the input and commenting it's different) I suppose \tracingall will reveal all...

@DavidCarlisle Oh, I think I found the issue. They are both unbalanced ifs, but regression-test will only complain about one Bad conditional the full log shows that indeed both are bad:

Bad conditionals: 1!

END-TEST-LOG

 )
(\end occurred when \ifcat on line 199 was incomplete)
(\end occurred when \ifcat on line 199 was incomplete)

Obviously regression-test miscounts :)

@Skillmonlikestopanswers.xyz ah. that's probably a feature, you can't really guarantee sensible processing after an error I'd guess.

@DavidCarlisle nah, regression-test's \END tests for \ifnum\currentiflevel>2, but somehow that magic number 2 seems wrong here.

However, good night, I'm going to bed :)

@Skillmonlikestopanswers.xyz good plan:-)

grr. always when tlgs is updated my ghostscript breaks ;-(

@UlrikeFischer Problem with ConText solved, moral of this, not configuring $TEXMFCNF, only brings problems :)