TeX, LaTeX and Friends

6:43 AM

@CarLaTeX Thanks!

@JosephWright Happy B-day and all dreams to come true!

20 hours ago, by Paulo Cereda

@JosephWright you are mean

@JosephWright Happy Birthday

@DavidCarlisle Thanks

@DavidCarlisle ignore @PauloCereda's comments:-)

@JosephWright Happy Birthday from me too!

7:07 AM

@JosephWright So \detokenize and \unexpanded behave differently wrt to #? Or was I misunderstanding your comment? Though such technicalities are beyond by perview.

@FaheemMitha Yes, the do: \detokenize will double them

@JosephWright And \unexpanded won't?

Is this question not a dupe, then?

@FaheemMitha In the same way, \detokenize will add spaces after macro names, which when passing to Lua could be important (as Lua doesn't skip those spaces)

@FaheemMitha I can't think of seeing it before, but there are a lot of questions

@JosephWright I see.

@JosephWright Yes, there are.

@FaheemMitha No, because its an anonymous \toks, and the latter has different rules for # handling

7:10 AM

@JosephWright Ok.

Though I was seeing the same doubling behavior for both, but perhaps in the case of \unexpanded something else was doing the doubling.

@FaheemMitha It must be: without context it's hard to say

@JosephWright The context is my example. As I mention there, using either \detokenize or \unexpanded gives exactly the same result for the Lua texio.write_nl output.

@FaheemMitha You've also got to deal with Lua escaping: in the \unexpanded case, remember it only applies 'once'

@JosephWright Sorry, I'm not sure what you mean. What only applies 'once'?

@FaheemMitha Well here it's more about escaping: with tracing on you see that using [[ ... ]] escaping, the # token stays doubled (standard TeX behaviour), but with \luaescapestring things get tided up. I meant more generally that once you \detokenize you ever have to worry about the catcodes, but if you \edef something twice, you need \unexpanded for both uses

7:26 AM

Oh, so \luaescapestring does some tidying up? People here seemed to have mixed feelings about it. But I guess it depends on the context.

@JosephWright @DavidCarlisle seen the mail?

@UlrikeFischer Yup

@FaheemMitha It 'knows' about # token treatment

@FaheemMitha As you'll have noticed, I favour using it over [[ ... ]] as it works correctly with cases like this, or if you had ]] inside your string, or ...

7:44 AM

@UlrikeFischer no was driving let me look..

8:00 AM

@JosephWright Actually, I hadn't noticed. But is "\luaescapestring{...}" a better default than [[...]] then, for passing args to Lua?

@FaheemMitha Like I say, what happens if ... contains ]]?

8:33 AM

@JosephWright Does that mean you do favor the \luaescapestring version, then?

@FaheemMitha Yes, because it makes no assumptions about the input beyond the necessity that it's a balanced text

@JosephWright Ok, I'll bear that in mind.

@FaheemMitha yes, or better, arrange that neither is needed and don't pass arbitrary fragments of tex through the \directlua interface.

I recall a discussion about this some time ago. Maybe a couple of months. But I came away with a rather confused impression of the best choice to make.

@FaheemMitha I guess this comes from me writing code that has to be robust

8:35 AM

@JosephWright Robust is good, of course.

@DavidCarlisle I'm thinking in for example \str_if_eq:nn(TF)

Though one cannot guard against all contingencies.

@JosephWright yes sure sometimes it's needed, especially in generic code like that, but even there while that should be robust if someone does it, if a user is string comparing random paragraphs of tex markup, then "don't do that" is probably the best answer.

@DavidCarlisle As far as I know, it is robust: we've found no cases that fail with the current code (compared to \pdfstrcmp, which does just have whatever TeX-the-program has grabbed as an argument string)

oh they just changed the css for the adverts again...

8:55 AM

@DavidCarlisle you probably can blame @barbarabeeton and her "bug"-report.

@UlrikeFischer that had occurred to me

9:09 AM

@DavidCarlisle I blocked them.

So PGF is using LuaTeX.

@DavidCarlisle ooh

@FaheemMitha Er, only when it's available: they have some stuff that's LuaTeX-only

9:27 AM

@Skillmon I was blocking them just in user css with display:none on some classes but they came back...

@DavidCarlisle I use uBlock Origin to block these things automatically.

@JosephWright Is it not always available?

@PauloCereda yes I was re-styling the sidebar anyway so it was simple to add a line to drop those segments, but if they change the css it breaks again

@FaheemMitha pgf works with xetex and pdftex so of course lua is not always available

@DavidCarlisle Since I am an idiot, I'd rather use automatic stuff. :)

@DavidCarlisle in case you are interested: addons.mozilla.org/pt-BR/firefox/addon/ublock-origin

@DavidCarlisle Oh, right.

9:32 AM

@PauloCereda Same here

@JosephWright Nice!

@JosephWright It's also amusing when people mention ads on Youtube videos. I am like, "Ads... on videos?"

@PauloCereda :)

Oh no

@PauloCereda of course I find the Portuguese description a lot better than the English one I would have used, thanks.

@DavidCarlisle oopsie, apologies

9:45 AM

@PauloCereda As you know I am fluent in all languages, so it was not an issue.

@DavidCarlisle ooh

9:57 AM

@bp2017 as for using \noindent generally you would expect that it is not needed as the surrounding enviornment would set up the required shape, \raggedright` or \centering etc. Note that these all work as expected if you use \usebox and have no effect at all if you use \copy.

10:21 AM

@DavidCarlisle @JosephWright we want to change the handling of missing chars in lualatex (currently they are simply dropped). Do you think it would be okay if the output changes to this:

For such a document:

\documentclass{article}
\usepackage{fontspec}
\setsansfont{Libertinus Serif}

\begin{document}

Latin Modern: x\SS y

Libertinus: \sffamily (^^^^278d)

\end{document}

@UlrikeFischer so long as not too many people are playing tikz' games with relying on \nullfont to drop characters...

@UlrikeFischer are you drawing that X or using the current font's U+FFFD character or??

@DavidCarlisle that's the notdef-character (that's why there is only a space with latin modern, it hasn't a sensible glyph here).

@UlrikeFischer ah OK so it's whatever the font says. seems OK with me....

@DavidCarlisle \nullfont is okay.

@UlrikeFischer yes it has no characters :-)

10:30 AM

@DavidCarlisle the main question is if someone who didn't realize that there are missing chars will complain if suddendly spaces and notdef glyphs appears. (One can disable it with a font feature).

@UlrikeFischer they will complain but they might complain less once you point out they were getting bad output before and didn't notice. Do you still get the missing character message in the log?

@DavidCarlisle yes: Missing character: There is no ➍ (U+278D) in font LibertinusSerif:mode=node. I will make it the default. If they complain too much I will blame you ;-).

@UlrikeFischer @JosephWright All graphics state operations (except transparency) are done in pure Postscript. No need to do it on the pdfmark level.

@UlrikeFischer blame Joseph (@PauloCereda said it was a good day to do that)

@DavidCarlisle ooh

10:41 AM

@AlexG the problem is the backend abstraction.

@PauloCereda I'm not mean, you see.

@UlrikeFischer Hm, what in particular?

@AlexG finding the right point where the backend level starts ;-). And which commands can be used to get the same effect for all backends (if it is possible). For bdc and /Properties I have a working model: github.com/latex3/pdfresources/blob/backendtest/experiments/…, and now I'm looking for something similar for ExtGState and ColorSpace.

@UlrikeFischer Ah, I see.

@AlexG I think for ExtGstate I will go through transparency commands and not try to offer an interface for all possible dictionary contents possible here - I didn't find any example using this.

@JosephWright did you see the lastkern discussion on the xetex list?

10:58 AM

@UlrikeFischer Yes; I see JK is on it

@UlrikeFischer I could give you a dvips example that sets a /Pattern colorspace, if that helps.

Hi .. I am trying to create a table like the above..

\documentclass{article}\begin{document}
\newcommand*{\myalign}[2]{\multicolumn{1}{#1}{#2}}
\begin{tabular}{lllllllllll}
\myalign{r}{a} & b & b &a & a &b & a &b &a & b \\

$q_0$ &$q_0$ &$q_0$ & $q_0$ & $q_0$ & $q_0$ &$q_0$ & $q_0$ & $q_0$ & $q_0$ & \\

& & & & & & & & & & \\

& & & & & & & & & & \\

\end{tabular}
\end{document}

I tried the above. based on a post on the site

The issue is, how do i specifically make the first row displaced?

@GermanShepherd several ways, simplest is to have twice as many columns and use && as separator so top row is just in even columns and others are just in odd columns

@UlrikeFischer what happens in hb versions of luatex which are doing similar passing entire runs to hb?

@DavidCarlisle Thanks..

11:16 AM

@DavidCarlisle sorry?

@AlexG yes, examples do help (the pgf code is in a part also an example but rather long ...).

11:28 AM

@UlrikeFischer Ok. The example defines a pattern, sets colorspace to /Pattern and uses the previously defined pattern as a "fill colour" of a rectangle:

\documentclass{standalone}

%define pattern prototype
\special{!
  <<
    /PatternType 1
    /PaintType 1
    /TilingType 1
    /BBox [0 0 10 10]
    /XStep 11
    /YStep 11
    /PaintProc {
      pop
      0 0 moveto
      10 0 lineto
      5 10 lineto
      closepath
  	  0.1 setlinewidth
      stroke
    } bind
  >> matrix makepattern /Triang exch def
}

\begin{document}

\makebox[200bp][l]{%
  \rule{0pt}{100bp}%
  \special{"
    /Pattern setcolorspace
    Triang setcolor
    0 0 moveto
    200 0 lineto

11:40 AM

@UlrikeFischer ignore me, I tested it myself

@DavidCarlisle now you made me curious ;-) What did you tested?

@UlrikeFischer just ran the posted example in harflatex

@DavidCarlisle you mean with the missing chars?

@UlrikeFischer no the lastkern thing on xetex list

@UlrikeFischer oh actually not sure I'm in harf mode by default in the version I have... (but OK anyway)

@DavidCarlisle Ah. I tried this yesterday and got the same as with lualatex and pdflatex (even with harf mode).

11:50 AM

@UlrikeFischer yes, seems that way:-)

@DavidCarlisle we probably will have to discuss the defaults for TUlmr.fd here. They don't set a mode and then harf is used, but perhaps mode=node is better.

@UlrikeFischer do you have an example where you get different output?

@DavidCarlisle until we change it, you can simply try the missing char example in a minimal document x\SS y. With luatex there is no space, with harftex + harfload there is one.

@DavidCarlisle but beside this, I don't think that I saw differences in the test files.

12:55 PM

@DavidCarlisle github.com/microsoft/MS-DOS

@PauloCereda Oh this is so cute :-)

@yo' :D

@PauloCereda yes, I saw that at the time I think. Finally you can install a real operating system, You may even be able to run emtex

@DavidCarlisle and emacs?

1:11 PM

@PauloCereda the only operating system that could be run in MSDOS was Windows until W98.

@yo' LOL

1:24 PM

@yo' oh, a Dirty Operating System

@PauloCereda could you give me suggestions on how I could draw this image above?

Are plans afoot to integrate Harfbuzz into LuaTeX? I'm not even sure what a text shaping library means.

@GermanShepherd TikZ and matrices?

@GermanShepherd no TikZ needed. Just some arrays and labelled arrows.

Q: How to install HarfTeX on TeXLive?

10

I've Installed TeX Live 2019 and downloaded HarfTeX. I'm using Visual Studio Code along with LatexWorkshop on Windows 10 and without any problem tex files compile with xelatex and lualatex. What to do now to make HarfTex work with it or with TexStudio? Here's my problem with lualatex without Har...

1:34 PM

@UlrikeFischer Yes, that's a fork.

@FaheemMitha the newest tugboat has an article from Khaled about it, so if you are a member you could read about it.

But is the plan to eventually merge it into LuaTeX?

@UlrikeFischer Not currently a member, no.

@FaheemMitha there is also an experimental luatex 1.11 which has harfbuzz too.

LWN has an article - lwn.net/Articles/741722

@UlrikeFischer Oh. So I guess the plan is to merge/integrate it, then.

So does this mean it can create fonts like Metapost/Metafont?

@GermanShepherd you can try something like the following:

\documentclass[border=3.14]{standalone}

\usepackage[]{amsmath}

\begin{document}
$
  \begin{array}[c]{|c|c|}
    \hline
    1&(\{s\},\text{white})\\
    \hline
    2&(\{\emptyset\},\text{white})\\
    \hline
    3&(\{\emptyset\},\text{white})\\
    \hline
  \end{array}
  \overset{b}{\rightarrow}_H
  \begin{array}[c]{|c|c|}
    \hline
    1&(\{s,g_1\},\text{white})\\
    \hline
    2&(\{g_1\},\text{white})\\
    \hline
    3&(\{\emptyset\},\text{white})\\
    \hline
  \end{array}
  \overset{a}{\rightarrow}_H

@GermanShepherd just noticed I put the empty sets in braces because of copy+pasting the first line... But I guess you get the point.

1:57 PM

@PauloCereda I was thinking of tables..

@Skillmon this helps a lot.. I 'll try this . Thank you@Skillmon

2:16 PM

@GermanShepherd little Bobby?

@HaraldHanche-Olsen LOL

@HaraldHanche-Olsen Im not sure I understand

@PauloCereda, since you are interested in automata theory ( I read your profile) , here is a question for you. Anbody else can also answer ;) while describing the executions/runs of a machine, lets say a finite state machine, it helps if we could animate it, to make the transition arrows move from one state to another on each input.

Have you ever tried something like this ?

@GermanShepherd Sorry, XKCD reference …

Her daughter is named Help I'm trapped in a driver's license factory.

@HaraldHanche-Olsen :P

@HaraldHanche-Olsen I believe I havent met you before. So, "Hi"

@GermanShepherd Hi. (I am currently on vacation, only dropping by now and then.)

2:32 PM

@HaraldHanche-Olsen If you do have any ideas on animating the arrows in a automata,on particular alphabets, do tell me. I am thinking of trying it out.

@GermanShepherd yes

@HaraldHanche-Olsen Enjoy your vacation !

@PauloCereda thats great. any examples or references perhaps?

\documentclass{article}
\usepackage{tikz}
\usetikzlibrary{positioning,automata}

\begin{document}

\newif\ifhighlight
\tikzset{
    highlight/.is if=highlight,
    highlight=false,
    step number/.initial=0
}

\pgfkeys{
    /tikz/id/.code={
        \pgfmathparse{ifthenelse(#1==\pgfkeysvalueof{/tikz/step number},"true","false")}
        \pgfkeys{/tikz/highlight=\pgfmathresult}
        \ifhighlight
            \tikzset{
                orange!60!red,
            }
        \fi
    }
}

\newcommand{\automaton}[1]{

@PauloCereda I was thinking of, doing it in beamer, with the arrows, showing up on each input.

\documentclass{article}
\usepackage{tikz}
\usetikzlibrary{positioning,automata}
\begin{document}

\begin{figure}[h]
\centering
\begin{tikzpicture}[shorten >=1pt,node distance=2cm,on grid,auto]
\node[state,initial] (s) {$s$};
\node[state,accepting] (g_1) [above right=of s] {$g_1$};
\node[state,accepting] (g_2) [below right=of s] {$g_2$};
\path[->]
(s) edge node {b} (g_1)
edge node [swap] {a} (g_2)
edge [loop above] node {a,b} ()
(g_1) edge [swap] node {a} (g_2)
(g_2) edge [loop right] node {b} ();

For example. on this, automata, for a input w1=ababb, i could show the run at each frame/step, possibly coloured like you have done.

I am unfamiliar with pgfkeys, one more thing I gotta learn.

3:20 PM

@PauloCereda I'm not sure it could run in dos2 but I certainly had emacs running in dos at the time (@yo')

@DavidCarlisle :)

@DavidCarlisle I never had emacs running

@GermanShepherd You may be interested in the overlay-beamer-styles library.

\documentclass{beamer}
\usepackage{tikz}
\usetikzlibrary{positioning,automata,overlay-beamer-styles}
\tikzset{orange on/.style={alt=<#1>{orange}{}}}
\begin{document}
\begin{frame}
\frametitle{Ducks}
\begin{tikzpicture}[shorten >=1pt,node distance=2cm,on grid,auto]
\node[state,initial] (s) {$s$};
\node[state,accepting] (g_1) [above right=of s] {$g_1$};
\node[state,accepting] (g_2) [below right=of s] {$g_2$};
\path[->]
(s) edge[orange on=1] node {b} (g_1)
edge[orange on=2] node [swap] {a} (g_2)
edge[orange on=3] [loop above] node {a,b} ()

@DavidCarlisle Uh, trying again from my laptop – the chat interface on the ipad was not too friendly … I nave had emacs on DOS, but instead, in those days, I had BRIEF: Basic Reconfigurable Interactive Editing Facility, sold by Underware. ;-) Its configuration language had Lispy syntax but C-like semantics.

@DavidCarlisle ok, this makes a second OS to be run over DOS, in line with Windoze.

3:35 PM

@HaraldHanche-Olsen I had demacs compiled with delorie.com/djgpp

3:54 PM

@DavidCarlisle Ah, I see it requires 80386 cpus. I think I my DOS experience was mostly (all?) in the 286 era.

@HaraldHanche-Olsen I stuck to my sun workstation at work and didn't have a machine at home until I could get one that ran emacs:-)

4:17 PM

@DavidCarlisle Heh. My first computer at home was a BBC microcomputer. Not much chance of running emacs on that. No, the 286 was at work. Only after we got some Apollo workstations could I run emacs. And so I have ever since, also on Sun machines and PCs running FreeBSD and linux.

4:29 PM

@marmot Thank you @marmot.. I will check out overlay-beamer-styles !

5:09 PM

Does anyone know what Plain-Tex prohibits the use of \newtoks inside a \def? I can get around it by invoking \csname newtoks\endsname, but am wondering if it will come back to bite me?

@StevenB.Segletes \outer

@DavidCarlisle Thank you. Looking up usage of \outer ...

@DavidCarlisle So, if I am reading p.206 of TeXbook correctly, \outer is the problem, not the solution.

@StevenB.Segletes \csname newtoks\endcsname avoids outer nonsense

@DavidCarlisle That is what I had discovered, and was just wondering if it will bite back in an unexpected way

@StevenB.Segletes full documentation on \outer:

\outer is the most useless, annoying, "feature" in the whole TeX language. — David Carlisle Sep 12 '18 at 8:27

5:24 PM

@DavidCarlisle LOL! That makes me feel better.

5:35 PM

@DavidCarlisle -- You may have that opinion, but it has saved my neck quite a few times, before LaTeX existed.

@barbarabeeton you claimed that before in this chat but the example that you gave (using \newsomething in a group if I recall) wasn't really a good example of \outer (there are no good examples of \outer:-) not just on modern machines, removing all uses of \outer from latex was one of the first things we did for 2e back in 1993.

6:15 PM

@AlanMunn uber-facts.com/15-signs-that-will-make-every-pun-lovers-day

@barbarabeeton ^^ :)

Alan Munn

@PauloCereda I thought I was in for a treatise on the gig economy. :) But these are really funny.

@AlanMunn :D

@DavidCarlisle -- Unfortunately, I've lost access to all the old work I could have used to get an example. But I really did have need for it. I do believe that none of the places it was a problem got into production. That would have been even worse.

@PauloCereda -- Aaaargh!

@PauloCereda last one is for Joseph:-)

@barbarabeeton see older discussion here:

May 22 at 19:47, by barbara beeton

@DavidCarlisle -- you apparently never had to develop code in plain tex for highly nested stuff that had to be counted, and accidentally put \newcount within a group, hence running out of hashsize.

Small piece of advice: Don't start 168 concurrent wget instances.

6:33 PM

@DavidCarlisle -- Well, yes, I know you don't like that one and understand why. But there are other situations where it's really convenient to have an error message pop up rather early in testing instead of lurking for some long time until you've forgotten the rationale behind some code. I have a similar objection to LaTeX making all \newcommands \long.

@barbarabeeton \long has some utility (and we switched the default to non-long in xparse) but \outer really was a bad idea, even if you find a real case where it caught a real error 9and I doubt that is possible, to be honest) it has caused tens of thousands of pointless errors in otherwise perfectly good code, that is a really bad tradeoff.

@DavidCarlisle ooh :)

@Skillmon uh-oh... What happened, mr. rabbit?

@DavidCarlisle -- Perhaps I've been brainwashed, but we shall have to agree to disagree.

@PauloCereda ^^^ so sad, no ducks to confort him!

@UlrikeFischer ooh a flamingo

6:40 PM

@PauloCereda Here is some Time-sensitive humor... It is only funny to people of a certain generation: imgur.com/gallery/gAFv5rK Not sure how to embed it

@StevenB.Segletes oh my

@PauloCereda I wanted to fetch all files from a link (a folder structure storing 168 files), and scripted it, without counting/knowing the number of instances started. Had a partial system freeze as the kernel had to swap out like everything...

@barbarabeeton well I wrote the \newcount that's in latex and it's not \outer so i win this argument:-)

@PauloCereda and because of the limited bandwidth some of the files timed out, now I have to restart the whole process and have wget -c checking whether the files are already completely downloaded..

@StevenB.Segletes -- And what generation might that be? As the spouse of someone who once had an aspiration to be a medievalist, all I can say is, I'm glad he never learned to ride a horse.

6:51 PM

@barbarabeeton The generation that were children when that crazy daredevil would get prime-time coverage so that America could watch him break most of the bones in his body on regular occasions. So exciting to a young boy! No slights intended for medievalists!

@StevenB.Segletes Oh wow, even in the middle ages they had \draw[-{Stealth[bend]}] (2,3) to[bend right=5] ++ (-0.5,0.1);, so the bending library is older than I thought!

For those too young to remember, en.wikipedia.org/wiki/Evel_Knievel

@StevenB.Segletes -- Maybe it's a male/female thing. I was more fascinated watching Philippe Petit crossing between two points on a high tightrope.

@barbarabeeton It was kind of like a poor man's moon launch

@StevenB.Segletes -- Oh, I'm quite old enough to have watched, but it's not something I would have been tempted to try on my Vespa.

7:00 PM

@barbarabeeton I think bike and skateboard manufacturers also got the idea around the same time that, if they could promote this sort of high-impact sport, they would sell a lot more equipment. Orthopedic doctors smiled, as well. Consider yourself wise beyond your years for avoiding it.

@StevenB.Segletes -- Well, when first learning to ride the Vespa, I tried to climb a chain-link fence. I decided right away that stunts were not my forte.

@barbarabeeton Prudence is a developed trait...youth is wasted on the young, as someone famous said.

7:29 PM

Hi.

I am getting odd error on stuff that used to work fine few years ago. I have the fonts, I checked with fc-match.

<argument> \LaTeX3 error:
            Invalid operation fp_to_decimal(inf)

This is on lines setting fonts for different languages that I switch manually:

\newfontfamily\cjkfont{WenQuanYi Zen Hei}[]
\newfontfamily\devanagarifont{Lohit Devanagari}[]
\newfontfamily\tamilfont{Lohit Tamil}[]
%\newfontfamily\cyrilicfont{CMU Serif}
\newfontfamily\cyrilicfont{PTSerif}[]

Why?

...before I submit to TeX.SE....

This is with LuaLaTeX.

7:45 PM

@wilx on all the fonts?

No. Just the one in the error.

Oh. Sorry. I omitted the error line. ...

<argument> \LaTeX3 error:
Invalid operation fp_to_decimal(inf)
l.83 \newfontfamily
\cyrilicfont{PTSerif}

so it's the tamil font that is in error I assume, what happens if you just load that one?

Let me see.

Yes. It seems to be the culprit. If I comment it out it gets past that.

Is it broken somehow?

@wilx fc-match wouldn't tell you if the font is available to luatex (it doesn't use fc-config unlike xetex)

@wilx I don't seem to have the font. We could blame @UlrikeFischer oh or @MarcelKrüger is here, we could blame him.

@wilx No idea, I just read the error message you posted.

OK. Thank you anyway. I did not understand it was the line before the PTSerif that was the issue. I can workaround this, I think.

7:58 PM

I do have the font but it doesn't find it under that name, this works for me

\documentclass{article}

\usepackage{fontspec}

\newfontfamily\tamilfont{Lohit-Tamil.ttf}[]

\begin{document}

\end{document}

@wilx the line break in the error message means that TeX had seen the \newfontfamily token on that line but had not read further so it hadn't executed that command or seen the PSserif argument yet, it's the usual issue that the previous line was looking ahead for an optional argument so had seen \newfontfamily but then put that back as it isn't [ so the errr message shows a line ahead of the real error

Ah. I see.

8:29 PM

Reading TUGboat ...

@JosephWright didn't anyone buy you a good book to read? :-)

@DavidCarlisle I did ...

@DavidCarlisle amazon.co.uk/dp/048645035X/…

@JosephWright is that the same kind of group theory as I'd mean,or some Chemistry notion of group?

@DavidCarlisle Same as you'd mean

@DavidCarlisle There are selection rules in the way light interacts with matter, and they can be understood using symmetry and groups

@JosephWright oh might be interesting then (if you skip over the chemistry bits:-)

8:38 PM

@DavidCarlisle Point groups not space groups :)

@DavidCarlisle Do chemists have groupies too? I thought they were for us mathematicians.

@DavidCarlisle The chemistry will be quite light, it's essentially a maths book

9:07 PM

@DavidCarlisle what did we do?

@UlrikeFischer see @wilx report above, does this look familiar (I couldn't reproduce, the font loads for me)

1 hour ago, by wilx

<argument> \LaTeX3 error:
Invalid operation fp_to_decimal(inf)
l.83 \newfontfamily
\cyrilicfont{PTSerif}

@DavidCarlisle one probably need the log file to see if some files are outdated.

egreg

9:35 PM

@JosephWright Maybe you can find an application to chemistry of the monster group

@egreg Too much moonshine?

@UlrikeFischer: Here is a test case: gist.github.com/wilx/94780b32b8f5bc9c48ae03aab73aa910

@UlrikeFischer And here is the log: gist.github.com/wilx/8927c2365a1052b215ce261e5f4d8fcc

@wilx where is the font?

@UlrikeFischer: github.com/google/fonts/tree/master/ofl/lohittamil

@egreg ^^^ (the differential d's and imaginary i's are typeset correctly, i.e. upright)

9:45 PM

@UlrikeFischer Probably the problem is that the font does not advertise a x-height and does not contain a x, so MatchLowercase has nothing to match. I remember fixing something like this a while ago and can't reproduce locally, so probably something is outdated.

@MarcelKrüger that's was my guess too when I saw the scale option, but the example works okay for me.

Actually, I am not sure which font version it was, maybe it was Lohit Tamil from this page: pagure.io/lohit

@marmot -- Different traditions in physics and math.

@barbarabeeton The author is on the math faculty (even chair)....

egreg

@marmot A physicist under cover

9:58 PM

@wilx with this font I get the error. I bet @MarcelKrüger is right and something it wrong with the fontdimen.

@wilx What are you trying to do anyway? So ask fontspec to scale the fonts such that lower case letters match (MatchLowercase) but you use a font without any lowercase letters. So what should that mean?

@egreg Hmmmh, in all the talks of world-famous mathematicians I attended recently the d's and i's were upright ... all physicists?

@MarcelKrüger This is just a part of larger document generated by Pandoc with custom preamble parts.

egreg

@marmot Renegades

Here: gist.github.com/wilx/530f2069d4f0de208de72cb514da6c7e

10:07 PM

@egreg They refer to themselves as mathematicians....

@marmot you refer to yourself as a marmot

@DavidCarlisle Yes, because I am a marmot.

@marmot -- This is very strange to me. I am checking with reputable authorities with long math publishing experience in Canada and the U.S.

@wilx That't why I wouldn't use \defaultfontfeatures. (Also because I mostly avoid fontspec, but that's another issue). You can reset the default font features directly before loading the font by adding \defaultfontfeatures{}, then restore the settings after loading the font with \defaultfontfeatures{Scale=MatchLowercase}

@MarcelKrüger well \the\fontdimen5\font` is zero. But the good news are that with the new missing-char code there is no error ;-). So with next version it should work. @wilx.

10:14 PM

I see. However, that bit is part of Pandoc's LaTeX template.

@UlrikeFischer Ubercool.

@barbarabeeton Is Cambridge reputable enough?

@UlrikeFischer "It should work" depends on your definition of "working". The entire font is scaled to the size of lower case letters.

@UlrikeFischer @MarcelKrüger thanks for looking at this.

@MarcelKrüger the output looks like this:

@marmot -- Cambridge U.K. or Massachusetts?

10:20 PM

@barbarabeeton Cambridge university press, so UK.

@wilx @MarcelKrüger but I think the script handling is wrong. With harf mode it looks like this:

The actual name in my PDF looks OK. The hooks connect nicely like in your second pic.

Also this leads to all the nice consequences of fonts with \fontdimen5=0pt, especially 1ex=0pt. This causes all kinds of trouble, for example tex.stackexchange.com/questions/488428/fandol-with-luatex

@marmot -- Certainly reputable, but U.K. I believe that usage is likely regional, so that's what I want to tie down. I never saw an upright "d" or "i" before maybe the late 1980s, and then it appeared in an ISO standard, which was devised by engineers, not pure mathematicians. I will admit that there is logic to the upright usage, but it simply didn't/doesn't appear in the mathematics publishing tradition.

@wilx The Lua code from that answer should fix your problem too even with current luaotfload versions by the way.

10:27 PM

@MarcelKrüger I had nearly forgotten this. I will add an issue so that it doesn't get forgotten again, imho fontdimen5 should never be zero.

@barbarabeeton For me it is the opposite. Let me pull out some random math lectures, web.ma.utexas.edu/users/a.debray/lecture_notes/m382d_notes.pdf, there the d's are upright. In fact, in my mind it meanwhile is such that I associate non-upright d's to sloppy, and thus more likely to be wrong. At least in the physics literature this is a valid rule of thumb, of course with exceptions.

@marmot -- I've just received a light-speed answer from a friend who has been one of the editors, involved in production, of the Canadian Math Journal. He agrees with me on the traditional usage. To quote from his answer: "[...] Nonetheless, we have followed the "forever" pattern and do all of them in math italic. This seems to be standard for Canadian publications including those in French." (cont'd)

(cont'd) Don Knuth uses italic. Proof is in the font information for cmmi, where "d" is the only letter without an italic correction, indicating that it is not (ordinarily) treated as a variable; if you want to use it as a variable next to another variable, you have to add a thinspace.

I should go to bed, I just read one of the question titles as "how to draw a dragon like this with tikz" ;-)

@UlrikeFischer -- Oh, but what if one wants to draw a dragon? I might very well want to do that.

@barbarabeeton thinking about it, we are really missing a number of dragons in the zoo.