7:03 AM
@UlrikeFischer,@MarcelKrüger Please check the latest DeviceN update

2 hours later…

9:00 AM
2
Not a palindrome, but...
@samcarter oh no

@PauloCereda It's more astonishing the green +9.9k!

@CarLaTeX ooh
@DavidCarlisle oh no

@PauloCereda simpler we could disolve you and distribute your remaining funds rep to willing takers

@DavidCarlisle oh no
@DavidCarlisle only if it's publicly announced in a mailing list
@JosephWright ^^

9:07 AM
@PauloCereda but we need to arrange it quickly, before dinner

@DavidCarlisle oh no
@DavidCarlisle I thought of calling the secretary, but then it would be two vulnerable ducks :)

9:21 AM
@JosephWright will do it in the afternoon. We are now waiting for the guide in the Petersdom ...

@UlrikeFischer OOH

10:02 AM
Hi folks. I was wondering if there was a standard way of dealing with the following. If I have a YAML config file (for use with LuaTeX) with a relative path in my TeX file which I pass to a library function, that library function is going to think the path is relative to its location, which will in general give the wrong location. The obvious thing to do is to complete to an absolute path. Are there other options? Yes, I can provide an example. And I could maybe post to the site.
A popular approach for TeX is to to put the file in a location it will search for. But not sure how that would work a YAML config file. Will it search in those locations for any missing file whatsoever?
E.g. under ~/texmf.

10:17 AM
@FaheemMitha presumably you are looking forthe file along the tex input path (although hard to tell from your description) in which case the fact that it's yaml makes no difference and you can place it anywhere luatex would find tex files

@DavidCarlisle I'll post a question with an example.
I've no idea where the system is looking for files.

@FaheemMitha well ~/texmf/tex for example

Completing to an absolute path name seems to be frowned on, for some reason.
@DavidCarlisle And subdirectories of that?
So even a Lua file will look for files in those directories?

@FaheemMitha as I say if you are using read or input then it's same as tex if you are calling kpathsea from Lua you can specify which search path it uses
@FaheemMitha it might (although you hadn't said you were calling from Lua, see my previous comment)

@DavidCarlisle Calling kpathsea from Lua? I don't know what that means. If I try to access a file by name in a Lua file, the question is whether it will search in those directories.
Anyway, I'll write an example.
This seems like an issue that will recur, though this is the first time I've encountered it for YAML.

10:28 AM
@FaheemMitha luatex manual sec 4.2.2

@DavidCarlisle Looking. Thank you.
@DavidCarlisle "4.2.2 Integration in the TDS ecosystem"?

10:45 AM
@FaheemMitha yes
4
@PauloCereda palindromes done right ^^

@FaheemMitha Basically all input functions will interpret relative paths as relative to the current working directory, so unless you do something weird they will not be looked up relative to some library.

@UlrikeFischer Is this your tour guide? sc02.alicdn.com/kf/…

@MarcelKrüger Is that the working directory to which the top level TeX file belongs? Anyway, again, will provide an example.

@samcarter no that's lunch after the visit

@FaheemMitha That's the directory you are in when invoking LuaTeX. Most likely that is the directory to which the top level TeX file belongs, but you could run LuaTeX from another path (this would probably break all relative paths in your document though, so it's usually not recommended). You could change the current directory from Lua though (which I also wouldn't recommend).

11:01 AM
@MarcelKrüger Ah, OK. Well, it doesn't seem to be doing that. So I guess an example is in order. Thanks.

@DavidCarlisle ... and we finally learn the true reason behind the English Reformation :)
@PauloCereda Ducks are good at math: youtube.com/watch?v=pG8KorwT_Pg

11:51 AM
(removed too)

@UlrikeFischer but your control of colour isn't as good as Joseph's

I just noticed that (using scrartcl), a call to \NewDocumentCommand worked without including xparse.

@FaheemMitha Latest LaTeX kernel includes ltcmd, which is the 'good' part of xparse

@JosephWright Did it not before?

1

If you know a recent LaTeX is available then you only need to explicitly use \usepackage{xparse} If you want to use one of the features that were trialled in the xparse package but were not included in the final version (ltcmd) that was incorpoated into the format. Notably the g argument type. B...

@FaheemMitha no (ltcmd didn't exist before)

11:57 AM
@DavidCarlisle OK, thank you.

12:29 PM
If you are a moderator, would you be okay or irritated if I flag a greetings comment?
Normally on stack they're removed, but there's an excessive number of those here that...

@user202729 People flag them, if they are old comments I usually delete them
@user202729 We have a different tradition: saying hello to new people seems OK
@user202729 The main site has a unique vibe ;)

12:43 PM
Do we have a general question for using macro in TikZ? See tex.stackexchange.com/questions/12091/… and its linked questions. (there are all the information already, but it's a little hard to find stuff)

@JosephWright When multiple Separations colorspaces use the same name, then the code does not respect which colorspace was specified. Especially: In m3color002.lvt, TwoUp is defined to use BarToneCMYK, but it ends up using BarToneCIELAB instead. (Probably not an issue in real documents, but maybe something we should add a warning for?)

@MarcelKrüger Hmm, probably need some tests, yes - feel free to adjust :)

@JosephWright At least for GhostScript, the code can easily be adapted to work with dvips too.

@MarcelKrüger Ah, cool: just update directly, doesn't need a PR?

@JosephWright I have to fix the tests first, then I'll push it to main.

12:56 PM
@MarcelKrüger Great
@MarcelKrüger I wonder if we could move to the non-backend part, but I guess this code doesn't apply to dvisvgm or similar, so 'no'

@JosephWright I tend to agree. While not PDF specific anymore, it's still about writing PDF or PS syntax, so it will not work with other backends and it feels like backend code.

Is there something in LaTeX for regex_replace, but use a function to expand the result? For instance I want to x-expand (or digest-expand, but this is probably harder) any \E{...} nested arbitrarily deep in some expression and substitute it back

Actually I can probably replace it with #something then push it into a function.

@user202729 There's a long tradition of doing expandable work without regexes in TeX - can you outline what you want to achieve
@user202729 The regex engine is impressive but it's inevitably slow compared to custom code

Not really, just curious
\expandafter is terrible
\edef is a little less terrible
\exp_args and \cs_generate_variant is the best so far

1:25 PM
Although generally speaking, I think that some things that is currently done in the expansion engine (mostly the math stuff) could be implemented much faster in another programming language

PGF/TikZ doing arithmetic by manipulating them as string is terrible, for instance

@user202729 Well yes, if you are not limited to TeX macros
@user202729 pgf has LuaTeX-only code that can for example replace the macro-based floating point work
@user202729 We (LaTeX team) have mused about adding primitives to expose some floating-point maths, but it's not likely to be worth it

In this recently posted example, it looks to me like the text inserted from the macro has more than usual whitespace to the left and right. Can anyone confirm that? And if so, what can one do about it?

@user202729 tex only has character tokens though, it has no numeric literals in the sense understood by programing languages so maipulating things always comes down to "strings of characters in the end"

@DavidCarlisle \count registers?

@FaheemMitha you are adding a space either side of whatever is returned by \directlua

1:32 PM
@DavidCarlisle Where am I adding the space?

@JosephWright yes but even there still aren't any integer literals \count0=123 might look like that but 123 is three tokens it's not a single token as it would be in ... most other things
@FaheemMitha an end of line is a space

@DavidCarlisle I meant that \count0 is itself an internal integer

@JosephWright I know it is, that's why I chose my words carefully:-)

@DavidCarlisle End of line in the macro?

@FaheemMitha yes anywhere, You have been using tex for long enough to see at least one of egreg's answers surely (90% of his rep comes from telling people to put a % at ends of lines)

1:36 PM
@DavidCarlisle 90%? Really?

@FaheemMitha at least:-)

@JosephWright Would you be interested in a matrix beamer room similar to the tikz/pgf room?

@samcarter There's a TikZ room?

@FaheemMitha ... and here's a guide how it works
12

Since a few days TUG has an official chat on the Matrix network. Below you can find some instructions on how to join the chat. Please let me know in the comments if anything is unclear. Matrix chat Matrix is an open, federated messaging platform. Like old-fashioned IRC channels, Matrix chats a...

@samcarter Ah, I see.

1:41 PM
@FaheemMitha I wouldn't use a binary lua module (and shell-escape) just to read a list of key:value pairs, shell-escape is a tool of last resort really. You could read it with a simple lua (or tex) loop, unless you really need a fully conformant yaml parser for all the edge cases of yaml syntax

@samcarter Enter the Matrix? Will I encounter Agent Smith? Should I be scared?

@FaheemMitha you must take the red pill :)

@DavidCarlisle Yes, the assumption is that one wants a full YAML parser. Or as full as Lua offers.

@DavidCarlisle Right (but that's why arithmetic in TeX can be expected to be (number of digits) times slower than that in C in that case)
... probably Lua will not be faster if that's the only overhead

@DavidCarlisle Actually, I'm not sure if -shell-escape is needed. I added that out of habit, really. Since one is not actually writing anything other than the usual PDF.

1:44 PM
@user202729 true but if you stick to the intended use of integer arithmetic (adding 1 to a section number every page or so) then it does it fast enough/

@samcarter Is the software running the Matrix free software? I could find out, but I'm lazy.

@FaheemMitha but it's not really good form to post code on the internet and ask people to run it with shell-escape enabled. It gives you access to their entire machine with whatever permissions they have. that code could do anything.

@DavidCarlisle I'll check and see if it works without -shell-escape.

@FaheemMitha from @HenriMenke's post: "Matrix is an open, federated messaging platform."

No, it seems to fail at finding the external Lua modules.
@samcarter I saw that. But that doesn't mean the software is free, does it?
I don't see why TeX restricts access to the external Lua modules by default. Possibly a security meaure. I wonder if it can be turned off.

1:49 PM
@FaheemMitha Further down the same post "Matrix is federated and it is also open-source"

@samcarter Ah. Thanks.

@FaheemMitha same reason it has shell-escape, you can make it default to unrestricted on your own installation but that can't be a general default (or good idea) a compiled lua module can do anything

@DavidCarlisle Well, it only calls two Lua packages, both of which are well known, though not part of Lua. And would normally be distribution packages. At least they are for Debian.

@FaheemMitha that does sure but you asked why compiled lua modules in general are disabled by default

I wonder if all of TeX Live has been audited for malicious code. It's a dangerous world out there.
@DavidCarlisle Ah yes, I see. Thank you for answering my question.
I'm reminded of this daily, because my email host sucks, and I get wagonloads of emails trying to infect my system. At least, I assume that's what they are trying to do.

1:54 PM
@FaheemMitha the code called from tex yes, needs a very rigourous review to get into the list allowed without shell escape.

@DavidCarlisle Oh.

@DavidCarlisle Anyway, that code will simply not work without the external YAML Lua module/library, at least. If you have suggestions on how to make it safer (reduce the number of paths?) let me know.
It would be good if Lua included more libraries by default, and then perhaps things would be safer, but it doesn't.
As far as I know, Debian has had no extreme examples of malicious code injection, though there was a kerfuffle some years ago. Someone added some code they shouldn't have.

2:19 PM
Is this ready now? Can I close that?
---> I wonder why the installer does not close automatically...

@cis Since it shows you a message which would be very hard to read if it would disappear immediately?

@cis Yes, you can

Ok, thx.

2 hours later…
4:23 PM
Had my old iMac 2011 opened and cleaned up for the first time today! Had to replace a cooler, now the machine is back (running Linux, of course). :)
3

2 hours later…

@PauloCereda ohh :)

@samcarter there's a free DLC coming too :)
@samcarter maybe it's my TUG 2020 talk in place :)

@JosephWright the deviceN code looks good. Even for dvips. What did you and Marcel do to get the object numbers?

ooh

@MarcelKrüger did you hear the talk about color fonts?

6:49 PM
@UlrikeFischer I don't think so. At which event was the talk?

@MarcelKrüger the pdf days. Thomas Merz said something about them.

Oh, I just found it. No, I missed it. Is there a recording?

@samcarter ooh

@MarcelKrüger theoretically yes. They sent me a message for the recording link, but in this one the actual link is missing, but I asked for a resent and will forward. I will also sent you the slides. It made me wonder about the technique we are using for color fonts. He suggested type3 but this is not what we are doing aren't we?

7:04 PM
@UlrikeFischer We mostly use virtual fonts instead, but I'm thinking about switching to Type 3 for some of them.

@MarcelKrüger I just tried with noto and it doesn't show any font at all, only images.

@UlrikeFischer Virtual fonts only exists in TeX, they disappear when on the backend. (One of the reasons for thinking about Type 3 instead)

7:21 PM
@DavidCarlisle oh no :)

@DavidCarlisle oh no
@UlrikeFischer oh no
@JosephWright ^^ help mr. duck

alta moda ^^^

7:37 PM
@UlrikeFischer Sebastian's old tee shirt

7:55 PM
@PauloCereda I still have my 2002 iMac G4. Haven't booted it in a long time.
@PauloCereda I really loved that particular design and wish it would come back somehow.
I've been spending a lot of time lately creating my own LaTeX3 examples and a lot of the programming philosophy is making sense now. I still have a few problems manipulating sequences and clists.
I can create correctly named functions with the programming layer and use them from the document layer.
Variable naming now makes sense to me too.

@LaTeXereXeTaL it could be that it's not that the scheme makes sense but rather we have corrupted you into accepting nonsense.

@DavidCarlisle Well I now read everything with an English, German, Italian, or Portuguese accent and that seems to help.

@LaTeXereXeTaL seems reasonable

@DavidCarlisle I'm retiring in December and I plan to spend many, many more hours working with it.

@LaTeXereXeTaL good luck with your retirement

8:10 PM
@DavidCarlisle I seriously want to do something to help the LaTeX Project in some capacity.
@DavidCarlisle I plan to continue evangelizing LaTeX use in introductory physics courses and to my academic friends in general who can't understand why using \left...\right everywhere is bad.

8:23 PM
Hello everyone,

May I know why the line

\setmathfont{Latin Modern Math}[version=bold,math-style=ISO,FakeBold = 3]

makes the following not compile

with error

Missing character: There is no 𝛽 (U+1D6FD) in font cmr10!
\documentclass{article}

\tracinglostchars=3 % Make it an error if a glyph is missing from the current font
\usepackage[english, bidi = basic, layout = lists]{babel}
\usepackage{fontspec} % For \defaultfontfeatures

\defaultfontfeatures{ Scale=MatchLowercase, Ligatures=TeX }

\babelprovide[import, onchar = ids fonts, mapdigits]{arabic}

\babelfont{rm}[Ligatures={Common,Rare}, Scale=1.0, Language=Default]{Latin Modern Roman}

\babelfont{sf}[Ligatures={Common,Rare}, Language=Default]{Latin Modern Sans}

8:51 PM
@Diaa I get multiple error before that. eg ! LaTeX3 Error: Invalid operation fp_to_decimal(inf)
@Diaa ah no sorry if I comment out the ScheherazadeNew references as I don't have that font then the error you show is the first
@Diaa a more reasonable test example is:
\documentclass{article}

\tracinglostchars=3 % Make it an error if a glyph is missing from the current font

\usepackage{fontspec} % For \defaultfontfeatures

\usepackage[bold-style=TeX]{unicode-math}
\setmathfont{Latin Modern Math}[version=bold,math-style=ISO]

\begin{document}

$\beta$

\end{document}
\documentclass{article}

\tracinglostchars=3 % Make it an error if a glyph is missing from the current font

\usepackage{unicode-math}
\setmathfont{Latin Modern Math}[version=bold]

\begin{document}

$\beta$

\end{document}

@DavidCarlisle you can even remove the bold-style option. What is unicode-math doing there?
@DavidCarlisle Oh I think I remember, for some reason if using a bold version you have to set the normal version too:
\setmathfont{Latin Modern Math}[version=normal]
\setmathfont{Latin Modern Math}[version=bold]`

9:11 PM
@DavidCarlisle @UlrikeFischer Thanks for your help

@UlrikeFischer sounds like a feature

@UlrikeFischer I knew I could blame you

9:41 PM
@DavidCarlisle ;-). there are some issues in unicode-math.

@UlrikeFischer latex-package-support

9:54 PM
@DavidCarlisle yes, something like that.

@JosephWright Do you have any objections about github.com/josephwright/beamer/pull/725 ?