TeX, LaTeX and Friends

« first day (4709 days earlier) ← previous day next day → last day (490 days later) »

mickep

07:13

@DavidCarlisle a secret, context uses latin modern 10 by default, even in 12pt or other sizes. (One can use design sizes, though)

Paulo Cereda

08:09

@AlanMunn ooh

Joseph Wright

08:19

@mickep Er ...

David Carlisle

@mickep you should use cmr10 and force the everyone use English plan.

Paulo Cereda

@mickep ooh dodos are very good at keeping secrets

@DavidCarlisle ay up me duck

David Carlisle

@PauloCereda breakfast

Paulo Cereda

@DavidCarlisle oh no

mickep

08:35

@JosephWright What? :)

@DavidCarlisle Oh, why not.

@PauloCereda The best!

Joseph Wright

@mickep Scaling rather than using design sizes :(

mickep

@JosephWright As indicated, there is a typescript also if you want to use the design sizes. But so far, I've not seen any real complaints.

Joseph Wright

@mickep Sure, most people won't notice, but it's the principle that bothers me: the font design has provided design sizes, one is supposed to use them 'as is'

mickep

@JosephWright I guess you are also against microtype then. :)

Paulo Cereda

Ducks like microtype

Not as much as koalas :)

Joseph Wright

08:44

@mickep Not really: microtype is about kerning, which is language-dependent in any case, whereas design size is about the shape of glyphs, and that's the core art of font design

mickep

@JosephWright No, microtype is also about expansion of the glyphs.

David Carlisle

08:58

@PauloCereda you eat koalas? (I prefer duck)

Paulo Cereda

@DavidCarlisle no :) and oh no :)

2 hours later…

samcarter

10:38

@DavidCarlisle Do they taste like cough drops?

mickep

11:10

@JosephWright There is also more to it. The Latin Modern Math font only comes in one size (with ssty1/2 design sized chapes, of course...).

Skillmon

@DavidCarlisle ^^^ Yes.

David Carlisle

@mickep can/does luametatex handle combining or variation selector sequences in math? getting some push back to my "VS1 sequences don't work in tex anyway" comment on using them for script variants...

mickep

11:29

@DavidCarlisle Not sure I get what you ask. What would be an example of that?

Ulrike Fischer

@DavidCarlisle do you know a font that actually has them?

Paulo Cereda

@UlrikeFischer ooh albatross

Ulrike Fischer

@PauloCereda öh. Can it detect glyphs from a variation selection?

David Carlisle

see w3c.github.io/xml-entities/U0FE00.html which says for example that U+2297 is ⊗ where the x touches the circle and U+2297 U+FE00 is ⊗︀ where allegedly it does not.

Paulo Cereda

@UlrikeFischer i've no idea of what that means, so I will quack :)

David Carlisle

11:44

@UlrikeFischer @mickep stix two math has them but only work in text:

\documentclass{article}

\begin{document}
$ $

\font\xc="[STIXTwoMath-Regular.otf]" \xc   ⊗ and ⊗︀

\bigskip

\textfont0=\xc
\textfont1=\xc
\textfont2=\xc
\textfont3=\xc

$  ⊗ and ⊗︀ $

\end{document}

@WillRobertson ^

mickep

11:59

@DavidCarlisle Not sure what I should test.

\setupbodyfont[stixtwo]

\startTEXpage[offset=1DK]
\im{a ⊗ b ⊗︀ c}

\im{\ee^{a ⊗ b ⊗︀ c}}

\stopTEXpage

David Carlisle

@mickep ooh it works:-)

mickep

(Do they have different sizes in script?)

David Carlisle

@mickep so we can blame @UlrikeFischer that it doesn't work in lualatex math I assume. (I know in principle it could be made to work in Lua, just wasn't sure if it was done anywhere)

@mickep in xetex for example the VS1 produces the white rimmed version of otimes in text as the pair is sent to harfbuzz together and harfbuzz handles the combination but in math it just tries to typeset each token and so VS1 produces no change in the output and you get Missing character: There is no ︀ (U+FE00) in font [STIXTwoMath-Regular.otf]!

Ulrike Fischer

12:22

@DavidCarlisle hm, with unicode-math I get Missing character: There is no ︀ (U+FE00) in font cmmi10!. Why cmmi?? Do one need to set a mathcode?

mickep

@DavidCarlisle Or we can say hooray to Hans who actually thinks of these kind of things. :)

David Carlisle

13:02

@UlrikeFischer you could set a mathcode but that would just change the font in the error, the problem is that in math it's not taking runs of characters but handling and spacing each separately so combining and variation characters are never applied to the preceding character they are just typeset as data individually so producing missing character warnings in most fonts, but not actually working anywhere. If you load the math font and use it as text it works

Ulrike Fischer

13:34

@DavidCarlisle yes, but basically it proves your point: there is not natural handling, one has to do an extra pass to replace the chars (and as far as I can see that is what luametatex is doing, there is e.g. a report_fonts("replacing %C by %C",char,variant) in the code).

2 hours later…

mickep

15:21

It is not a luametatex thing, it is a context thing. (Also works in mark IV)

Alan Munn

16:12

TeXShop users: (if there are any here) have you encountered problems with the PDF window being blank? This is an intermittent problem for me and I can't figure out how to reliably reproduce it.

Werner

For those around, SE is interested in an experiment. More available here:

Q: Do we want to volunteer to test the lowered voting privilege?

There is a post in Meta.SE that is seeking some volunteer sites to test an adjustment of voting privilege (currently set at 15 to upvote and 125 to downvote). There are a few changes, but the highlights are. What would change for sites participating in this test? There are three main privilege c...

3 hours later…

Ulrike Fischer

19:31

@JosephWright are issues in latex3 closed when merged or when uploaded?

Ulrich Diez

Why does \@writechkpt write to \@partaux and not to \@auxout ? (Sometimes I write checkopints with documents that consist of a single file. E.g., with the following, you can toggle \onlytoctrue/\onlytocfalse for choosing if the entire document is to be typeset or only the table of contents is to be typeset.)

\documentclass{article}

\newif\ifonlytoc
%\onlytoctrue

\ifonlytoc\nofiles\fi

\makeatletter
\newcommand\writecheckpoint[1]{{\let\@partaux\@auxout\@writeckpt{#1}}}%
\newcommand\setfromcheckpoint[1]{\@nameuse{cp@#1}}%
\makeatother

% Rest of the preamble

\begin{document}
\ifonlytoc\csname fi\endcsname\setfromcheckpoint{BeforeToc}\tableofcontents\end{document}\fi

\section{One} Bla \section{Two} Bla \section{Three} Bla
\section{Four} Bla \section{Five} Bla \section{Six} Bla
\newpage
\writecheckpoint{BeforeToc}\tableofcontents

Joseph Wright

20:13

@UlrikeFischer when we merge

3 hours later…

cfr

23:40

@DavidCarlisle tex.stackexchange.com/questions/696612/… I thought the OP had a 'user base'?

David Carlisle

@cfr probably all the users are vim users, so it doesn't really matter.

cfr

@DavidCarlisle ???????????????????????????????????????????????

David Carlisle

@cfr don't expect too much coherent conversation in this chat:-)

cfr

@DavidCarlisle Surely you don't use emacs or something?

David Carlisle

@cfr is there an alternative editor?

cfr

23:48

@DavidCarlisle emacs is supposed to be an editor? They disguise it well.

Why do I get ! Use of \??? doesn't match its definition.? Where does the \??? come from?

David Carlisle

@cfr It comes from ~~Bruno~~ the expl3 layer in the format, expl3's expandable error message where you are in a context when undefined command or doesn't match are the only possible errors you can throw.

cfr

@DavidCarlisle Oh.

David Carlisle

@cfr texdoc.org/serve/interface3/0#subsection.11.4.2

cfr

@DavidCarlisle The thing is, it almost always gives me something meaningful right after, so it seems odd.

David Carlisle

@cfr yes clever code, it uses an intentionally invalid command to smuggle a readable error message out to the user.

« first day (4709 days earlier) ← previous day next day → last day (490 days later) »

all rooms