12:01 AM
@DavidCarlisle what I took away from all of you is that you agree that mathematical documents must be accessible to the greatest number of existing illnesses, is this correct?

@manooooh you don't seem to answer any of the points raised, why look at more documents? Which do you think a screen reader will read correctly relaci ́on or relación
@manooooh no sometimes production methods mean it's infeasible, but authors should care
@manooooh for some uses this is anyway a legal requirement.

@DavidCarlisle I do understand the meaning of "screen reader". I guess he was not interested in adding those packages because anyone is able to read that PDF with the accents, it is not necessary (according to what he says) that the accented characters can be copied correctly, that he prefers to say his time to do other things

223

The two packages address different problems. inputenc allows the user to input accented characters directly from the keyboard; fontenc is oriented to output, that is, what fonts to use for printing characters. The two packages are not connected, though it is best to call fontenc first and the...

@manooooh Whoever says that they don't need inputenc when not using UTF-8 is wrong.

@DavidCarlisle I do not understand you

@manooooh anyone here means anyone reading the text by eye not having it read to them or translated to braille or ...
@manooooh I mean sometimes you have to use a particular method to make a document, perhaps post it to a blog site (or stackexchaneg) or have it printed by a publisher and you have no control, so sometimes you can not control the format of the final file or how accessible it is, but that is not the same as saying it is OK not to care about these things.

12:07 AM
@manooooh but it means that the pdf can not be searched easily either. Do you only look at pdf or do you some copy and paste and search in a pdf? In the second case you should care about correct accents as they have an impact on how you use a pdf too.

@egreg but I guess he was not referring to the fact that he uses inputenc, but that "there is no problem in adding \usepackage[T1]{fontenc}"

@manooooh so why not do that?

@DavidCarlisle I am sorry I still do not understand you
Do you say that authors should worry about brailling their documents?
@DavidCarlisle he writes all his documents on his own, he does not have a style guide from anywhere, he simply uses common sense

@manooooh you said "anyone can read the pdf" and that discounts whetever proportion of the population use automatic screen readers or other accessibility tools as the accented letters only look like accented letters in teh internal pdf they are unaccented letters with dijointed accents (in a private encoding) overlayed
@manooooh they should have concern for that yes. They don't need to know anything about the details just take basic steps to make their document accessible. If you can cut and paste the text as text then it can be read out loud or converted to braille automatically. If you can not cut and paste the text then most likely it can not.

@UlrikeFischer I agree with you, and he also agrees with that, but he does not have enough time for it. Even, as he provides the link of the docs to people, if we add the possibility of accepting accents then the links will not be the same as before, which would cause a huge break of all links to his documents, that is, a disaster. Manually changing each link is crazy
@DavidCarlisle I do not know. He also does not agree that fixing his metadata in all the docs is more productive than doing anything else, because he will always find something better to do than correct the metadata of his documents

12:14 AM
@manooooh saying there is no time is wrong and not an excuse. It takes no more time to typeset a document in T1 encoding than OT1, it produces better hyphenation, searchable text and far better accessibility
@manooooh people say that about there documents, and they say that about not putting ramps to allow access to buildings and lots of other things, that's why theer are laws forcing people to care...

@DavidCarlisle any screen reader can show the accents for the reader. He says it's okay to have a style, but he does not agree to follow anyone's rules, within what is reasonable

which points out that posting an inaccessible document to a goverment website is "probably breaking the law" that of course is the UK law but most other countries (certainly all of the EU and USA) have similar rules

@DavidCarlisle he is a professor in addition to other things. Have you seen the number of pages that a single document had, and the figures that he uses? He has published many other books, so it would take a lot of time to compile all the documents again (according to him) no one or very few notice whether the accents are copied well or badly. With respect to accessibility, it would be completely broken, since those books with accents will now break the access hyperlinks

@manooooh "any screen reader can show the accents for the reader" what do you mean by that? How would a screen reader know that the symbols in tex-specific encoding are accents at all?
@manooooh sorry that makes no sense, I stop here:-)

@DavidCarlisle he is not interested in following any rule, he acts under what he considers to be reasonable. Keep in mind that these documents undergo changes in terms of corrections of operations, formulas, etc., but he is not interested in copying accents correctly do not fix their metadata since that would be the last in your list of things to do
@DavidCarlisle the motto is "while the user reads the accents, everything will be fine", everything else is a finer spin
@DavidCarlisle what is in discussion is whether spending time on finely spinning is necessary in a math, when the important thing is that he uses his time to publish mathematical documents
In fact he uses accents, but arranging accents and correcting metadata are last on the list of useful things to do

12:30 AM
@manooooh switching to T1 encoding costs nothing, that is done in 2 seconds. There is no excuse not to do it.

@UlrikeFischer the author agrees to do so (at some point)
@DavidCarlisle Don't use \hbox in a LaTeX document, and what about \mbox? Can \mbox be used in a LaTeX document?
@DavidCarlisle why should I pay attention to that page?

1:10 AM
@daleif that is contradictory. If a function could be identified with its domain or with its codomain, then the absolute value would be the same function as f(x)=x^2, since both have the same domain and the same codomain
I correct myself. I should have said that | |={(x,y) in R^2: (x>=0 wedge y=x) vee (x<0 wedge y=-x)} and not |x|={x in R: (x>=0 wedge x) vee (x<0 wedge -x)}

4 hours later…
5:27 AM
@marmot I think you should go here and add some explainations (and possibly an answer as well)! ;-)

@JouleV You could say : 33 is special. It is the smallest integer divisible by 33. ;-)

@marmot Hahaa ... it is very funny ;-) but the OP will be very confused though :))
I would say 33 is the sum of the first four positive factorials: 33 = 1 + 2 + 6 + 24.

@JouleV To be honest, 33 was the first thing that came to my mind, also because it is easier to type the same number twice (especially when you are born with claws). There is no mystery other than that I was looking for some element which seemed to have the maximum width.
@JouleV That's a nice observation! Do you know Ramanujan?
Srinivasa Ramanujan FRS (; listen ; 22 December 1887 – 26 April 1920) was an Indian mathematician who lived during the British Rule in India. Though he had almost no formal training in pure mathematics, he made substantial contributions to mathematical analysis, number theory, infinite series, and continued fractions, including solutions to mathematical problems considered to be unsolvable. Ramanujan initially developed his own mathematical research in isolation: "He tried to interest the leading professional mathematicians in his work, but failed for the most part. What he had to show them was...

6:00 AM
@DavidCarlisle I agree

6:19 AM
Latex experts: I am not sure if this is a latex problem or not, thought to ask first. I compile a PDF file using lulatex which has one \includegraphics on a image file (the image file is also a pdf file). When I open the generated PDF file in local PDF reader (Adobe), the image shows up OK. When I open the same PDF file in chrome build-in reader, the image inside the PDF do not show up. There is white area where the image supposed to be. Could this be latex issue or chrome PDF reader issue?

@Nasser Have you checked that the pdf is OK? Please see also tex.stackexchange.com/q/436054/121799, of course I do not know if it is the same problem.

@marmot Ok, will run ps2pdf on it and see. Thanks.
I run an online tool to validate the PDF and it says it is OK.
File foo.pdf
Compliance pdf1.5
Result Document validated successfully.
Details
Validating file "foo.pdf" for conformance level pdf1.5

The document does conform to the PDF 1.5 standard.

Done.
The latex is just this

\documentclass{article}
\usepackage{graphicx}
\begin{document}
\includegraphics[width=0.6\textwidth]{p1}
\end{document}
This is very strange. both p1.pdf and foo.pdf are OK when I look at them locally. It is only chrome pdf reader which does not show the image inside. May be a bug in chrome.

@Nasser test if instead of writing p1 you write p1.[file extension] e.g. p1.jpg

@manooooh The code is above. Again, no error compiling with lualatex.

@Nasser I know you said that you have no errors, but try adding this, since you have a problem with the file once it is compiled

6:33 AM

@Nasser how can you show a PDF file using includegraphics? I think you have to convert the PDF into eps or another kind of extension
30

Have a look at the pdfpages package \documentclass{article} \usepackage{pdfpages} \begin{document} \includepdf[<options>]{<file>} Your Text \end{document} Edit Miguel suggested graphicx to include a graphic which is correct, the difference between graphicx and pdfpages is that graphic...

@manooooh includegraphics can read images in PDF format? I do this all the time

@Nasser I do not know O.o, I am learning with you haha

1 hour later…
7:39 AM
@manooooh pdflatex can not include eps images but can include pdf, you have that backwards

7:58 AM
@DavidCarlisle got my mail?

@UlrikeFischer er let me look...
@UlrikeFischer ah. OK:-)

@DavidCarlisle hello sir
@DavidCarlisle I found more information about mode and line break in section 815 of TeX the program.
@DavidCarlisle so thank you very much and good day. I'll back when confused (very soon :-))

8:52 AM
Can anyone provide a link that refers how to change the text font of TextField (hyperref package) to Computer Modern i.e. the standard text font of LaTeX?

9:10 AM
@manooooh that's rather difficult, as a pdf embeds only subsets of fonts, so what should happen if your textfield wants to show a char you haven't included? If you have the otf-versions as external system font you can try tex.stackexchange.com/questions/36630/…, but this is not very portable.

@UlrikeFischer thanks for the comment! I only need to write 2/4 numerical chars in differents TextField. I do not know if I have otf-versions or not, I am using LaTeX->PDF output, if that is important
If it is possible to do but it is rather difficult we could ask a question

@manooooh Why do you care about the font in the textfield?

@UlrikeFischer because it is part of the document, and I consider that TextField must have the same font as LaTeX's font

9:31 AM
@manooooh Well as I said: if the font is there as a system font, you can use it see also tex.stackexchange.com/a/469012/2388. But if you move the pdf to some other PC it will probably break if the font is missing there. I don't think that you can embed a font for this as a resources, but I don't intend to investigate more. That's not important enough for me to care about it.

@UlrikeFischer that answer does not provide the name of the font I am looking for. I thought that any PDF viewer supported the LaTeX standard font in TextField but it seems that no. Thanks!

9:46 AM
7

@samcarter Welcome back!!!
2

@AlexG Thanks for the welcome back! The good news is that I found several new tikzlings in the last month: github.com/samcarter/tikzlings

@samcarter welcome! And when will you upload to ctan? I could make use of some of them.

@UlrikeFischer I'm planing a new release soon - there is one new tikzlings in the makings, as soon as this is finished I'll push them to ctan.

10:29 AM
@samcarter Welcome back !

@AndréC Merci!

10:48 AM
@samcarter I will add my voice to the “welcome back” chorus.

11:28 AM
@HaraldHanche-Olsen Thanks also to you for the nice welcome back!

12:01 PM
@samcarter I'm disappointed you didn't use the break to re-implement them all in picture mode:(

12:11 PM
@DavidCarlisle I thought that one picduck would be good enough as proof of concept for chat.stackexchange.com/transcript/message/48991931#48991931

@samcarter Welcome back! :D
% Pupils %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\cat@part@draw[black] (-0.1720, 1.8071) .. controls (-0.1820, 1.7827) and (-0.1820, 1.7584) .. (-0.1720, 1.7341) .. controls (-0.1620, 1.7584) and (-0.1620, 1.7827) .. (-0.1720, 1.8071) -- cycle;
\cat@part@draw[black] ( 0.1720, 1.8071) .. controls ( 0.1820, 1.7827) and ( 0.1820, 1.7584) .. ( 0.1720, 1.7341) .. controls ( 0.1620, 1.7584) and ( 0.1620, 1.7827) .. ( 0.1720, 1.8071) -- cycle;
%

@PhelypeOleinik Oh, very good idea!
@PhelypeOleinik One could even make the width of the pupils adjustable

@samcarter :D
@samcarter Oh, it would be nice :)
@samcarter They need to see the tikzmice when it's dark ;)

12:35 PM
@samcarter @marmot is probably feeling aggrieved that he is still stuck as tikz and the ducks can upgrade to picture mode.

@PhelypeOleinik github.com/samcarter/tikzlings/commit/… (not yet adjustable, but I'll have a look tonight)

@samcarter <3

1:07 PM
4
@PauloCereda ^^^

@samcarter Try this: pastebin.com/rSccxANp

1:27 PM
@samcarter much tastier after the change.

1:58 PM
@egreg @DavidCarlisle Imagine if one day you don't answer a single question on TeX.SE, how much gain in reputation do you expect when you come back on the next day?

@JouleV I don't post as much as I did (tend to average ~100 a day not ~200+ these days) but it used to be not uncommon to be rep-capped at 200 before I looked in the morning.

@DavidCarlisle I see that Gonzalo Medina hasn't been on the site since 8 August 2017 and since then he has gained 50k rep :D.
He gets about 80-100 each day.
You have more rep than him so I think the number is not 80-100 :)

@JouleV yes I was just going to mention Gonzalo as an example
@JouleV I think I average less than 100 for such "old" points these days

2:21 PM
Hi, I'm currently testing some LuaTeX font stuff and need to debug some Adobe Problems. Could someone using Windows and Acrobat Reader send me a screenshot of Acrobat Reader rendering this document?

@MarcelKrüger sure. What do you want to see? some details or the overall impression?

@UlrikeFischer Thanks. I am interested in the overall impression with special focus on the third page.

@MarcelKrüger mail sent with the third page.

@UlrikeFischer Thank you very much.

@PhelypeOleinik Thanks a lot for the code! I'll have a closer look tonight.

2:39 PM
@MarcelKrüger ^^ (acrobat pro 9 it seems on this machine, not reader) ...

@samcarter Hi!

@DavidCarlisle Thanks, is the Firefox example with some Plugin or the built-in PDF viewer?

@marmot Hi! Did you see the tikzlings "dog" answer? (don't worry, it is marmot safe :)

@samcarter You're welcome :-)

@MarcelKrüger built in one (firefox nightly)
@MarcelKrüger edge's built in viewer also only gets the top of the brackets, Chrome shows the whole bracket

3:25 PM
@samcarter -- It's really nice to see you back! Thanks for the travelogue.

3:41 PM
Hello @DavidCarlisle and hello @ll

Quick question: how does TeX pick an editor to open when I type e after an error? It opened some editor and I had to kill the terminal to quit it :/

@barbarabeeton Thanks :) How are you going to celebrate tomorrow's Proofreading Day daysoftheyear.com/days/proofreading-day ?
@PhelypeOleinik If you have to kill the terminal just to quite an editor this must be vi :)

@samcarter Shush, I don't want people to know I have it in my computer ;-)

@samcarter -- Oh, my goodness! That had totally escaped my notice!. Maybe another new TUGboat article will turn up. (But I know some people who have a more active interest, and I shall share.)

@PhelypeOleinik understandable :) (sorry, too late to edit my previous message)

3:50 PM
Just popping in quickly to say “Welcome back” to @samcarter.

@barbarabeeton Maybe we can ask @DavidCarlisle to send you a few teh for this occasion
@Circumscribe Thanks also to you! It's nice to see you all again!

(I noticed because the scale factor question got an upvote.)
(Gotta go though.)

@samcarter -- Hmmm. I manage to type that myself much too often!

@samcarter No worries :)

3:54 PM
@Circumscribe Let nobody disturb your circles :) Bye!

\documentclass{article}
\begin{document}
\begin{picture}(0,0)
\put(0, 0)t
\put(10,0)h
\put(5, 0)e
\end{picture}
\end{document}
3
@samcarter ^

@barbarabeeton see the message above :)
@DavidCarlisle Oh, good use of picture mode :)

@DavidCarlisle -- Hmmmmpf.

@samcarter Yes. Cats and marmots are friends. ;-)

@samcarter you got mail ...

4:15 PM
@UlrikeFischer C...a...n...n...o...t r...e...p...l...y l...a...u...g...h...i...n...g t...o...o h...a...r...d

4:44 PM
@JouleV Rep cap, probably. :-)

5:17 PM
@samcarter ooh
@samcarter welcome back. :)

@PauloCereda Your timing is fantastic -- I was just quoting your code: tex.stackexchange.com/a/478248/36296

@samcarter ooh
@samcarter OH MY

Does someone know shy this isn't working as expected?
\documentclass{article}

\begin{document}

\showhyphens{programa}
% [] \OT1/cmr/m/n/10 pro-grama

\hyphenation{pro-gra-ma}

\showhyphens{programa}
% [] \OT1/cmr/m/n/10 pro-grama

\end{document}

@samcarter Welcome back! Gorgeous!

I was expecting the latter \showhyphens to show [] \OT1/cmr/m/n/10 pro-gra-ma

5:31 PM
@CarLaTeX Thanks :)

@DavidCarlisle t‮he
(look carefully; it's not what it seems to be)

5:46 PM
@PhelypeOleinik you are forgetting the hyphenmins:
\documentclass{article}

\begin{document}
\lefthyphenmin=1
\righthyphenmin=1

\showhyphens{programa}
% [] \OT1/cmr/m/n/10 pro-gra-ma

\hyphenation{prog-ra-ma}

\showhyphens{programa}
% [] \OT1/cmr/m/n/10 prog-ra-ma

\end{document}

@UlrikeFischer Ooh, those. I knew it was something kind of stupid :/
@UlrikeFischer Danke :-)

6:10 PM
@samcarter Well, this gives you the perfect answer for the dog question. [Hint: use \let\praiiredog\marmot .[ ;-)

Is there a way to define a \newcolumntype with a macro as the name, i.e., \newcolumntype{\foo}{...}?

@AlanMunn sure, simply do it.

@UlrikeFischer Really?
\documentclass{article}
\usepackage{array}
\def\coltype{C}
\newcolumntype{\coltype}{c}
\begin{document}
\begin{tabular}{C}
foo
\end{tabular}
\end{document}

@AlanMunn ah, no, that doesn't work. You then need to use \coltype in the tabular.

6:27 PM
That can be fixed by adding enough \expandafters:
\documentclass{article}
\usepackage{array}
\def\coltype{C}
\expandafter\newcolumntype\expandafter{\coltype}{c}
\begin{document}
\begin{tabular}{C}
foo\\foobar
\end{tabular}
\end{document}

@AlanMunn \newcolumntype takes a token as argument, so X or \coltype or \blub is all fine for it. So you need to expand it if you want the "inner" token.

@MarcelKrüger Perfect. That's what I was looking for.

6:47 PM

7:04 PM
@DavidCarlisle ‮We could break the chat again...
Bah it just reverted the text...

7:29 PM
Hello all
I have a quick question that I think is probably faster to figure out in chat than posting a full question
In essence, I have a page that has a couple graphics separated by flexible white-space. Sometimes all three graphics can fit on one page, other times only the first two should fit. In the latter case, I'd expect that the first two graphics stay on the first page, distributed even vertically, and the third graphic would be on the next page. However, my code always keeps all three graphics together. I'm of the impression it's due to \vspace creating an unbreakable vbox, but I don't know how else to handle what I want without manually doing it (which is undesirable; the end goal is templates t

@DavidCarlisle, @UlrikeFischer, @egreg Let's see if texdev.net/2019/03/07/… gets us lots of b-type xparse questions :)

Apologies, I forgot the MD for code sections

7:48 PM

8:02 PM
Okay, I think I got something that works now, but I feel like it's kind of messy
i replaced all the \vfill lines with

\vspace{0.1in plus 1fil}
\vspace{0in plus -1fil}
And I get the desired result. But that seems a bit messy to me. Is there anything cleaner than that?

@JosephWright That's really quite nice.

8:40 PM
@JosephWright I have to review a lot of answers!

@Shamtam sounds very odd. I don't see why there should be no page break at the \vfill. But without a complete example, it is difficult to say more. Better ask a proper question on the main site.

8:59 PM
@DavidCarlisle ‮Aye.

9:10 PM
Why is fontspec so slow with LuaTeX? It seems to take almost 10 seconds to load while with XeTeX there is no delay at all.

@AlanMunn I have no problems. Can you show an example?

\documentclass{article}
\usepackage{fontspec}
\setmainfont{Linux Libertine O}
\setsansfont{Linux Biolinum O}
\setmonofont[Scale=MatchLowercase]{Linux Libertine Mono O}
\begin{document}
\end{document}
This takes about 4 seconds on my machine. (Is there an accurate way to time it?)

Yes, say time lualatex <filename>
On my machine it takes 7 secs.
But on the second run "only" 2 secs.
Probably because first the fonts needed to be created/loaded/whatever.

@marmot It takes 6 seconds on mine, with no subsequent speedup.

@AlanMunn Do you compile in a terminal or do you use TeXShop. (I never use TeXShop.)

9:19 PM
@AlanMunn which tex system?

@marmot I'm using TeXShop, but the times are from running in the terminal.
@UlrikeFischer TL 2018 on a Mac.

@AlanMunn is it up-to-date?

@UlrikeFischer Reasonably. I'll update now to see if there are any recent changes.

@AlanMunn An l3kernel update could change something, Joseph improved the loading of the unicode-data a lot.

@UlrikeFischer Would that only affect LuaTeX? With XeTeX the same document takes 1.2 seconds.
@UlrikeFischer After updating, still no change: 6s for LuaTeX, 1.2s for XeTeX.
@UlrikeFischer As in your answer here I wonder if it's something to do with the font name search procedure.

9:27 PM
@AlanMunn I could speed it up considerably by compiling
\documentclass{article}
\begin{document}
\end{document}
instead. It has the same output. ;-)

@marmot But I prefer my empty documents in Libertine. :)

@AlanMunn do you have the same difference without libertine?

@UlrikeFischer No, if I don't load Libertine, I get 1.5s for LuaLaTeX.

@AlanMunn and if you load only the main font?

@UlrikeFischer 3 seconds

9:40 PM
@AlanMunn is this also much slower than xelatex?

@AlanMunn where is the font loaded from? Did you try to load it by file name instead of font name?

@UlrikeFischer If I load a system font (from /Library/Fonts) it's faster. Libertine is in the texmf tree.

@Shamtam that is just \vspace{0.1in} that is you are replacing stretchy glue by a fixed length. But impossible to say what is happening without a real example.

@AlanMunn What about other fonts in the texmf, e.g. TeX Gyre Termes. Are there also difficult when font names are used?

9:47 PM
@AlanMunn when I were a lad, people used to be happy to have a page rendered in under quarter of an hour. What's the rush? :-)
2

@UlrikeFischer Seems about the same with Termes. 1.9s for LuaTeX.

@UlrikeFischer Yes, you're right. I was just too lazy to put together an MWE. I've posted it here: tex.stackexchange.com/questions/478309/…

@AlanMunn what means "the same"? also much slower than with xelatex?

@UlrikeFischer For one font, the difference seems not huge (2s LuaTeX vs 1.3s XeTeX) but with more fonts, it gets longer it seems.

@AlanMunn well small difference are to be expected. The main question is if there is something odd with libertine or with font name search on your system. But I can't test it. Perhaps @MarcelKrüger has some idea.

10:29 PM
@UlrikeFischer With the TeX Gyre fonts, loading three fonts as in the original example is relatively much faster. 2.3s vs 6s for Libertine (LuaLaTeX). XeLaTeX is 1.3s.

@UlrikeFischer I currently have no idea, especially after the first run the name lookup should be supplied directly from the lookup cache, so 6 seconds look like a long time. @AlanMunn Could you show us your lookup cache? It is a file named luaotfload-lookup-cache.lua most likely located in ~/.texlive2018/texmf-var/luatex-cache/generic/names/ or $YOUR TEX LIVE PATH$/texmf-var/luatex-cache/generic/names/.

@barbarabeeton congratulations on your election! can you stop brexit?

@AlanMunn sounds like a libertine specific problem.

@DavidCarlisle hmmm a long long time ago
@DavidCarlisle ooh

@MarcelKrüger Here it is:
return {
["AdobeCaslonPro/BI#bi#655361"]={ "/Library/Fonts/ACaslonPro-BoldItalic.otf", false },

10:37 PM
@PauloCereda what you broke this time? tex.stackexchange.com/questions/478314/…

@MarcelKrüger Which points me to places where I didn't realize I had fonts... :) I usually install them into /Library/Fonts rather than ~/Library/Fonts so that they're available to all users on my machine.

@DavidCarlisle oh no

@AlanMunn didn't you say that the libertine is in the texmf?

@AlanMunn This whole thing sounds similar to github.com/u-fischer/luaotfload/issues/40 Do the files linlibertine-*.lua in .../luatex-cache/generic/fonts/otl/ change if you compile the test document?

@UlrikeFischer Well I thought it was.

10:47 PM
@AlanMunn @UlrikeFischer never trust the users.

@UlrikeFischer @MarcelKrüger As an experiment, I removed all the fonts from ~/Library/Fonts and now the document won't compile at all with either XeLaTeX or LuaLaTeX using font names.
@DavidCarlisle Oi

@AlanMunn you rang?

@AlanMunn well xelatex won't use font names if the fonts are only in the texmf (unless you adapted the fontconfig). lualatex should normally use them, but perhaps you need to recreate the database (this naturally assumes that you have the fonts there ....).

@UlrikeFischer Well I do have them there. This I do know :) I've rebuilt the database and now the LuaTeX version runs. It now takes 2.3 s for three fonts, so that's a big improvement.
@MarcelKrüger @UlrikeFischer I need to go right now. But this discussion has been quite helpful.

@AlanMunn sounds as if Marcel's guess that the double fonts were the problem was right.

11:05 PM
@JosephWright time to update unicode-data before tl2019? Unicode 12.0 adds 554 characters, for a total of 137,929 characters. These additions include 4 new scripts, for a total of 150 scripts, as well as ... ?
@yo' @yo' have you seen the new U+1FA80 character?

@DavidCarlisle ooh

11:34 PM
@DavidCarlisle ;-). You got mail ;-)

@UlrikeFischer bottom left is best

@DavidCarlisle yes I think that's one main event of the gathering ;-).
@DavidCarlisle I hope you agree with the title ...