TeX, LaTeX and Friends

« first day (4370 days earlier) ← previous day next day → last day (859 days later) »

Joseph Wright

06:02

@UlrikeFischer All use intarray data, but the LuaTeX implementation doesn't use fontdimen's under the surface

@MarcelKrüger Ah, right

1 hour later…

Joseph Wright

07:15

@DavidCarlisle, @UlrikeFischer Have realised adding an optional argument to \MakeUppercase is tricky, as we currently have

\cs_new_protected_nopar:cpn{MakeUppercase~}{\text_uppercase:n}
\cs_set_nopar:Npx\MakeUppercase#1{\use:c{MakeUppercase~}{#1}}

so it's not obvious how to do it ...

I guess ltcmd, grab the optional argument expandably?

David Carlisle

07:27

@JosephWright since it's \protected is expandability needed here? We could always give direct two arg access to the l3 code if it is needed

Ulrike Fischer

08:05

@MarcelKrüger the error message for undefined commands is a bit curious, it always reports \undefined:

! Undefined control sequence \undefined .
<line 1.106>
    \blub

Joseph Wright

08:24

@DavidCarlisle We have to grab the mandatory argument and wrap in braces in an expansion context

David Carlisle

@JosephWright ah yes sorry, top level is not protected...

Joseph Wright

@DavidCarlisle :)

@DavidCarlisle I think I'm OK with \NewExpandableDocumentCommand \MakeUppercase then passing the results

Joseph Wright

09:01

Er, no, that won't work

3 hours later…

Joseph Wright

11:51

@DavidCarlisle Good plan ... trying now

3 hours later…

Paulo Cereda

15:03

@JosephWright follow this account: twitter.com/SoVeryBritish

Joseph Wright

15:30

@DavidCarlisle github.com/latex3/latex2e/pull/936 :)

Ulrike Fischer

15:41

@JosephWright why \reserves@a and not \reserved@a?

Joseph Wright

@UlrikeFischer Can't spell

@UlrikeFischer There's a more pressing issue with test failures in tools, but I may not be able to look properly until tomorrow

Ulrike Fischer

@JosephWright which one failed?

Joseph Wright

@UlrikeFischer tl2e1, tnfss1, tlb-varioref-00n

David Carlisle

16:27

@JosephWright thanks I'll look this evening

Ulrike Fischer

17:25

@JosephWright it breaks in the header (pagestyle headings) because of the \quad:

\documentclass{article}
\begin{document}
\MakeUppercase{xx \quad xx}
\end{document}

Joseph Wright

@UlrikeFischer That’s very odd: I wonder what’s triggering that

@UlrikeFischer what I don’t get is why it shows up now

Ulrike Fischer

@JosephWright with an optional argument it compiles: \MakeUppercase[lang=de]{xx \quad xx}

Joseph Wright

17:42

@UlrikeFischer I must have the argument forwarding wrong. I’ll work on it.

@UlrikeFischer Thanks

Ulrike Fischer

@JosephWright with the optional argument \reserves@a ->de in the trace, and without

\reserves@a #1->\cs_generate_variant:cn {text_\str_lowercase:n {#1}case:nn}{V}\
cs_new_protected:cpx {Make#1case\c_space_tl \c_space_tl \c_space_tl }[##1]##2{\
keys_set:nn {__kernel}{##1}\exp_not:c {text_\str_lowercase:n {#1}case:Vn}\exp_n
ot:N \reserves@a {##2}}\ExpandArgs {cnx}\NewExpandableDocumentCommand {Make#1ca
se}{O{}+m}{\exp_not:c {Make#1case\c_space_tl \c_space_tl \c_space_tl }[####1]{#
###2}}

I think you must initialize the key correctly.

mickep

18:18

@barbarabeeton (or anyone else): Do you know why \pm sits on the baseline in computer modern but is shifted down in latin modern?

mickep

18:34

Another (to me) curious thing: If I open cmmi10.pfb in fontforge, some symbols are repeated in more than one slot (see upcoming picture, where there are three occurences of \psi marked). I see no difference. Does anyone know why? Or is fontforge tricking me somehow?

barbara beeton

@mickep -- I question that \pm is sitting on the baseline. The entire "composite" should be centered vertically on the math axis. Set \pm next to plus; the horizontal bars shouldn't line up.

mickep

@barbarabeeton ^ That is compiled with pdflatex.

@barbarabeeton That is latin modern math.

wipet

\pm in Latin Modern looks better, it is exactly plus with minus added. Computer modern has bad position of plus in the \pm symbol (shifted up).

mickep

@wipet I'm fine with the lowered one as well, but it was new to me that it was changed.

David Carlisle

@UlrikeFischer (@JosephWright) doesn't work for me without the language argument. More generally I wonder if we should drop the \protected it's hard to see what advantage it has if we do everything even [] grabbing via expansion

barbara beeton

18:57

@mickep -- Well that's not what Knuth intended. I believe (without checking) that the minus has the same "depth" as the plus.

mickep

@barbarabeeton Yes, looks the same (I did not check numbers)

David Carlisle

19:15

@JosephWright ! LaTeX Error: The key '__kernel/langg' is unknown and is being ignored. are these all in a common kernel key pool, does that mean other stray keys will be accepted once more commands have kv options?

Joseph Wright

20:15

@DavidCarlisle That is to be clarified ;)

David Carlisle

@JosephWright but apart from not handling the default [] case currently I think it should basically all work.

Joseph Wright

@UlrikeFischer Fixed

@DavidCarlisle I was imaging that longer-term we might want bcp47 or lang set globally by babel or the kernel, and then locally re-set

@DavidCarlisle Should I hope now be sorted: @UlrikeFischer spotted the issue

David Carlisle

@JosephWright ok let me pull and rebuild...

\documentclass{article}

\begin{document}

\MakeUppercase{xyz}

\end{document}

@JosephWright I still get

! Argument of \reserves@a has an extra }.

Joseph Wright

@DavidCarlisle Try now

David Carlisle

@JosephWright better:-)

bzm3r

21:05

Is there a document or manual that explains how one would go about implementing TeX? I expected TeXbook to be it, but it seems TeXbook is more about using TeX than implementing it?

(also, would this be a good question for TeX.se? I couldn't find something relevant by searching, but maybe I didn't use the right keywords.)

Alan Munn

@bzm3r What do you mean by "implementing TeX"? Rewriting the source?

David Carlisle

@bzm3r tex-the-program is the published book sources or texdoc tex.pdf

@bzm3r most implementations use (at the core) Knuth's original source (tex.web) rather than re-implementing the algorithms from some specification

bzm3r

@AlanMunn that's right.

@DavidCarlisle hmm, I am having trouble parsing this sentence, could you clarify?

ah: tug.ctan.org/systems/knuth/dist/tex/tex.web

David Carlisle

@bzm3r I would normally interpret a question "how to implement some algorithm" to be asking how to take a high level specification, and coding it in some programming laguage, not taking an existing give source code and compiling it

Alan Munn

@bzm3r This might also be helpful.

bzm3r

21:15

@DavidCarlisle yes, I agree, and that is what I meant too

David Carlisle

@bzm3r yes but you probably have the generated tex.pdf on your machine already

@bzm3r few have done that. miktex, texlive both compile Knuth's web (Pascal) source via a custom "web2c" compile chain that convert the web to C then compiles

bzm3r

hm, no, i can't find tex.pdf on my machine

David Carlisle

@bzm3r texdoc tex.pdf would work with a full texlive

bzm3r

ah, Alan's link explains how I can build it,and also provides a link

right! cool

@AlanMunn thanks, that's quite an interesting resource

« first day (4370 days earlier) ← previous day next day → last day (859 days later) »

all rooms