TeX, LaTeX and Friends

Joseph Wright

9:13 AM

@PauloCereda /quack

Paulo Cereda

9:25 AM

@JosephWright quack

David Carlisle

10:43 AM

@PauloCereda /breakfast

Paulo Cereda

@DavidCarlisle oh no

samcarter

11:25 AM

@Skillmon I usually remember the correct one for chat, but then make some other mistake, like missing a space after it. However instead of noticing this mistake, I think that I misremembered the syntax and try the other one which of course also does not work :)

@Skillmon already did chat.stackexchange.com/transcript/message/60263911#60263911 (In the mean time, I offered a bounty, that's why I'm now below again.)

Skillmon

11:41 AM

@samcarter you had one on SO when I wrote that comment :P

@samcarter at this point this is fighting wind mills, I gave up...

samcarter

@Skillmon ups, missed that one :( ... @DavidCarlisle would have complained anyway about the improper positioning of the comma :)

Paulo Cereda

4

ooh a Don Quijote rabbit

samcarter

@Skillmon As long as this has not yet reached the tag pages, it's not that bad :)

@PauloCereda :)

Skillmon

@PauloCereda LOVELY!

Paulo Cereda

@Skillmon <3

Skillmon

11:48 AM

(and reminds me of some children's book with animals behaving like knights and stuff in some abbey)

Henri Menke

12:04 PM

@Skillmon Since pgfkeys is littered with \csname what we'd need is an efficient emulation of \begincsname for non-LuaTeX engines.

\ifdefined\begincsname\else
  \def\begincsname#1\endcsname{\ifcsname #1\endcsname\csname #1\expandafter\endcsname\fi}
\fi

This is the simplest possible code but doesn't support nesting of csnames.

And it already requires two expansions to get to the desired result.

@JosephWright You have experience with backporting primitives. Do you think it is realistic to backport \begincsname (and while we're at maybe also \lastnamedcs) to other engines?

Skillmon

@HenriMenke actually 3 (so 2 more than \begincsname). But steps of expansion != speed (at least not necessarily)

Henri Menke

@Skillmon It's not so much about speed than more about breaking every consumer that relies on a certain number of \expandafters.

Skillmon

@HenriMenke I remember begging for \lastnamedcs as well (which I find more useful than \begincsname)

@HenriMenke yes, that's understandable, I was just commenting because of "efficient" :)

Henri Menke

@Skillmon But \lastnamedcs is really only a performance optimization, right?

@Skillmon You're right, "efficient" was definitely the wrong term. I guess I meant "compatible".

Skillmon

@HenriMenke yes, plus you don't have to forward stuff in some places. Whereas with \begincsname it might expand to nothing, which can break code picking up the newly built control sequence, I prefer my \csname\ifcsname construct over that.

(and that would be the, imho, most compatible replacement for the \csname thingies throughout the code)

@HenriMenke It's impossible to define something that takes just a single step like \begincsname and behaves like it. I'd argue you should not try, but instead abstract from it in both cases, something like:

\long\def\pgfkeys@csname#1{\csname\ifcsname #1\endcsname #1\else relax\fi\endcsname}
\ifdefined\begincsname
  \long\def\pgfkeys@usecs#1{\begincsname#1\endcsname}
\else
  \let\pgfkeys@usecs\pgfkeys@csname
\fi

That would always take two steps of expansion, and if you just want to use the cs without having to forward it you could use \pgfkeys@usecs (which then could be more efficient in LuaTeX)

Henri Menke

12:22 PM

@Skillmon Oh, but that's even worse I think. The results will be different on pdfTeX and LuaTeX.

Skillmon

@HenriMenke that's why if you depend on the result being the same would use \pgfkeys@csname and if you don't depend on the result being always one token use \pgfkeys@usecs.

Henri Menke

Also I think \par is forbidden inside \csname so no need for \long.

Skillmon

@HenriMenke And that's why I dislike \begincsname :)

@HenriMenke \pgfkeys@csname{\detokenize{\par}}

@HenriMenke \pgfkeys@csname{pgfkeys@handletokens@\pgfkeys@count{#1}}

(I can think of infinitely many variants having a \par inside the argument without the \par ending up in \csname :))

(also, with \expanded available in all major engines: \long\def\pgfkeys@usecs#1{\expanded{\ifcsname #1\endcsname\unexpanded\expandafter{\csname #1\endcsname}\fi}} yields the same as \begincsname)

(what would be cool was if the engines would provide a way to have a macro start an expansion context without needing two steps of expansion, in LuaTeX this is possible because you can define new pseudo-primitives via the Lua interfaces)

Henri Menke

12:38 PM

@Skillmon Thank you for providing all these valuable insights. I have adapted my fix accordingly. Could you please review github.com/pgf-tikz/pgf/pull/1132 because I've now listed you as a coauthor.

Skillmon

@HenriMenke Oh no, not again :P Yes, I'll take a look.

David Carlisle

@JosephWright ^ discussion of csname expansion possibilities seems to be the topic of the month:-)

Skillmon

@DavidCarlisle because of the file-parse code?

David Carlisle

@Skillmon keys (parallel team discussion:-)

Skillmon

@DavidCarlisle I have no inside into team discussions (yet -- and probably never will)

David Carlisle

12:50 PM

@Skillmon you don't miss a lot especially these days we usually open a github issue if there is anything that is worth discussing in public which reminds me I need to open an issue about braces killing the option clash code

Skillmon

@HenriMenke should I also add comments for other stuff I'd change or just review the changes?

Joseph Wright

@Skillmon would you want to?

@DavidCarlisle indeed

Skillmon

(\pgfkeysaddvalue could use some changes, it uses \toks1 with a local assignment and also should get groups around the whole thing to not accidentally have side effects around other things using \toks0 and the other scratch register, the assignment to the key macro is global anyway)

Joseph Wright

@HenriMenke these ones look quite easy

Skillmon

@JosephWright probably a bad idea :) I don't have enough free time to really be useful to the team, and I might have the wrong coding style in general (too much head-through-the-wall being expandable everywhere and using too many specialised auxiliaries along the way -- which might be fine for performance critical stuff). I guess the modus operandi with the occasional PR is better for both sides right now.

Joseph Wright

1:02 PM

@Skillmon sounds good to me:)

@Skillmon My aim is that the team provide cute ideas, and as you say we find some’ friends’ who look at specific areas they are interested in

Henri Menke

@Skillmon If you see some low-hanging fruit you can fix it in an extra PR or just send a patch on Matrix.

Skillmon

@HenriMenke never mind the group thingy for \pgfkeysaddvalue, there is a group that's just using braces instead of \begingroup...\endgroup

Lunch for now, until later!

Marcel Krüger

1:45 PM

@PauloCereda Are you islanders already aware that texdoc.org's TLS certificate has been revoked?

TeXnician

@MarcelKrüger The island is not hosting it. Only the underlying software is IoT. Hosting and running is @StefanKottwitz.

Marcel Krüger

@TeXnician Thanks.

TeXnician

@MarcelKrüger Thanks for reporting. It seems strange enough. Usually, these Let's Encrypt certificates should be auto-renewed by Caddy (at least in our recommended Docker setup). But maybe @StefanKottwitz can provide further insight.

Marcel Krüger

@TeXnician They normally get auto renewed when they expire, but a bunch of certificates had to be revoked earlier due to a Let's encrypt bug.

TeXnician

@MarcelKrüger Interesting, haven't been aware of that.

cis

6:13 PM

So Ms. Fischer, I guess I have to hack your whole chess package.
I think I can create something that, given a starting situation AND ONLY INPUT OF THE MOVES, can display any of the following game situations.

I would have wished somewhere that the package could do something like this by default. However....

@UlrikeFischer

john

Hello everyone! I'm working with this template github.com/fmarotta/kaobook it's really nice, but has wide margins (I need that since I take note on margins) and I have a long equation which I can't split. Anyone has some suggestion on how to write a long equation?

Rmano

@john it is almost impossible to blindly split an equation over lines because semantic matters there. You can have a look at tex.stackexchange.com/questions/3782/…

john

@Rmano I don't need to split the equation, but rather scale it on the page width and not on column width

Ulrike Fischer

@cis I don't understand your problem, if your moves are legal chess moves then you can use skak or xskak to set a start position and to play them.

Rmano

@john So, do you want the equation going into the margin? In that case, the solution is dependent on the class you're using and on a lot of details, so really you should prepare a short example and post it on the main site.

john

6:24 PM

@Rmano exactly. I thought that maybe there was an easy way to solve this. I'll try post on main site. Thank you!

Rmano

tex.stackexchange.com/questions/176443/… and/or tex.stackexchange.com/questions/156877/…

cis

My chess puzzle is not really a normal chessgame.
Maybe, I can show you the result and you can say me, whether it could have been easier then.

Rmano

But my suggestion is that if an equation is overlong, the best possible thing is to split it semantically: define sub-blocks to have a name and use them. This will enhance readability a lot.

john

@Rmano ok with the second link i managed

Yes usually i split them, but this looses readability by splitting

barbara beeton

6:55 PM

@john -- Scaling a long equation to a size that's not easily readable decreases the ability tor the reader to understand it, often more so than splitting it. That's why it's not easy to "make it smaller" automatically -- it requires thought by someone who knows what it means.

We had a quite nice blizzard here yesterday. Today the sun is out and it's very bright, but very cold. This post from someone at a local TV station is pretty typical: twitter.com/KaylaFishTV/status/1487835796237398019

Ulrike Fischer

8:55 PM

6

@CarLaTeX ^^^ your doing

CarLaTeX

@UlrikeFischer Happy to have had part in it!

samcarter

@UlrikeFischer Nice palindrome! Congrats!

David Carlisle

9:12 PM

@UlrikeFischer three ducks, reminds me of the dinner we just had

barbara beeton

@UlrikeFischer -- Very nice! Even flipped over 180 degrees, still a palindrome, although the comma suddenly rises.

2

barbara beeton

11:58 PM

@DavidCarlisle -- I'm confused by this question about longtable: tex.stackexchange.com/q/632106 In addition to the question about different numbers of lines on the first and second pages, I fail to see why the pages are being set in portrait mode rather than in the requested landscape. longtable doesn't appear to do anything that should affect the orientation (and hence the page length).

David Carlisle

@barbarabeeton will look later but there is space before the table so line count change not that surprising

Transcript for

$TeX - LaTeX$ TeX, LaTeX and Friends