12:49 AM
@Canageek No problem. If only Boris Johnson would do the same ....

1:00 AM
@DavidCarlisle After all that, my forest turned out too wide and I couldn't use the coffins after all. However, my other document presumably breaks with 2020, so I'll need it there. Thanks!

@cfr coffins in general will work OK, coffins with lists in them less so, sorry about that, it managed to pass all our tests except the one Joseph just added which is a copy of your test file so it stays fixed:-)

@PauloCereda You don't like them? I think they're fantastic. One of the most useful l3 thingies.

@cfr he's just scared of what's inside

@DavidCarlisle When you say they'll work OK, but not with lists, do you mean they won't work with lists after the fix? Or they just won't work with lists until a/the fix?
My other document also has lists. That's why I know it'll break: it is exactly the same thing: lists on the left and forests on the right.
\NewDocumentCommand\twotoo{m+m+mo}{%
\medskip\par\noindent
\SetVerticalCoffin\lcoffin{.5\textwidth}{\linewidth=\hsize#2}% workaround for TL2020 bug from David Carlisle
\SetVerticalCoffin\fcoffin{.48\textwidth}{{\centering #3\par}\IfValueT{#4}{\noindent[#4]}}%
\JoinCoffins\lcoffin[r,vc]\fcoffin[l,vc](.02\textwidth,0pt)%
\SetHorizontalCoffin\fcoffin{\textsf{Question #1}}%
\JoinCoffins\lcoffin[t,l]\fcoffin[t,l](0pt,\baselineskip)%
\TypesetCoffin\lcoffin[T,l]\medskip\par%
}
#2 always is a list in this document :(.

@cfr no they will work fine after the fix, the code didn't change at all, it's just that when it got preloaded into the format \linewidth isn't defined so it made a version of the definition (suitable for plain tex) that just set \hsize so adding the linewidth setting should give exactly the old behaviour (and we should get a patch level 3 format out next week that adjusts teh test for "latex" so it works again)

1:09 AM
@DavidCarlisle Oh, good. I thought that's what you meant, but the 'will' had me worried.

@cfr sorry I meant if you use the existing broken definition in other documents (like every one in our test suite apparently) that have coffins without lists then they will work OK. \linewidth is (mostly) just used in lists
@cfr I am sure this comment would read better as workaround for TL2020 bug, fix from David Carlisle :-)

@DavidCarlisle I use them all over, though mostly the expl3 syntax, but you're right. I don't put lists in them all the time. (But I use lists for formalising arguments a lot.)
@DavidCarlisle I meant the workaround ;).
@DavidCarlisle It's only so I know who to blame if it breaks.

@cfr Ulrike

@DavidCarlisle New version:
\NewDocumentCommand\twotoo{m+m+mo}{%
\medskip\par\noindent
\SetVerticalCoffin\lcoffin{.5\textwidth}{\linewidth=\hsize#2}% workaround from David Carlisle for TL2020 bug
\SetVerticalCoffin\fcoffin{.48\textwidth}{{\centering #3\par}\IfValueT{#4}{\noindent[#4]}}%
\JoinCoffins\lcoffin[r,vc]\fcoffin[l,vc](.02\textwidth,0pt)%
\SetHorizontalCoffin\fcoffin{\textsf{Question #1}}%
\JoinCoffins\lcoffin[t,l]\fcoffin[t,l](0pt,\baselineskip)%
\TypesetCoffin\lcoffin[T,l]\medskip\par%
}

:-) good night

1:16 AM
@DavidCarlisle Nos da!
@DavidCarlisle Should have stuck to commenting my files bilingually.

5 hours later…
6:01 AM
@HaraldHanche-Olsen All days are organised TV station free. :) We watch what we want, when we want. :p
@HaraldHanche-Olsen for us, that is. :)
@barbarabeeton :D It's got pickled everything. :p Whey was a very readily available liquid for preserving food, so you can get pickled slátur (haggis) in two varieties (liver and blood); pickled meat in various guises (flanks, both straight up and pressed, and "head-jam"), as well as the piece de resistance: pickled rams testicles. :p
@cfr Bore da! :p

2 hours later…
7:59 AM
@cfr cwac! Oh I do, it's because there's something in the works about them. :)
ooh and quack
@samcarter_preparing_for_xmas ooh

baa! or... bwaa!?

@Plergux ooh welsh sheep... or welsheep? :)

8:18 AM
@PauloCereda :p quite possibly. I was encouraging the kid to learn Danish (just started learning it at school) by getting him into Duolingo. And I thought I might as well do one myself, to remember reminding him. So I picked Welsh. :p I can say "Bore da, Paulo! Sut dych chi?" but not much more :p
"Hmmm.... I should probably redo that Zipf plot, it's a bit blurry. I wonder if I could use that pgfplots thing... Holy hell, the manual is longer than my thesis!!!" \faints

@Plergux ooh Cymru
@DavidCarlisle meanwhile, % arara: context :)

9:08 AM
@AlexG I wonder if you have an idea how to set the pdf version in the tex file when using dvips?

9:31 AM
@PauloCereda tell that to @yo'

9:41 AM
@UlrikeFischer got rid of zzz.log so you see the real errors now in the context example overleaf.com/read/gqkyhmftnrrw

10:02 AM
@UlrikeFischer looks like we failed to get that to work last time we looked at that:
Jan 19 at 13:00, by David Carlisle
@UlrikeFischer but I can't get latex-dvips-ps2pdf to work without specifying 2.0 on the commandline for ps2pdf I thought <</CompatibilityLevel 2.0>> setdistillerparams would work but it's accepted by gs but seems to be ignored (and the helpful manual for setdistillerparams says "PostScript language programs that use these operators must not assume that any particular key is present in the dictionary returned by currentdistillerparams, or that setdistillerparams has any particular side effects."

@UlrikeFischer I ran a test with ps2pdfwr and putting, e. g., <</CompatibilityLevel 1.2>> setdistillerparams into the PS file. setdistillerparams seems to be ignored, as @David told. Also, any commandline setting via -dCompatibilityLevel=... greater than 1.7 is ignored.

@DavidCarlisle The IoT TL images have support for all the TeX tooling available.

38 mins ago, by David Carlisle
@PauloCereda tell that to @yo'

@DavidCarlisle IoT will always be the underdog in the TeX world, it seems

@PauloCereda all you need to do is set up a server using the iot images, and get 6 million people to have accounts on that server
@PauloCereda not so much an underdog, more an island.
@yo' I have context working at overleaf now, no worse than the documented platex support, just a three line latexmkrc file overleaf.com/read/gqkyhmftnrrw
@Skillmon wow regex, so massive overkill:-) I hate to think how many more expansions that takes rather than the one expansion that I suggested:-)

10:26 AM
@DavidCarlisle I know, but this way he can use \mythetitle everywhere without having to locally redefine \\  everytime.
@DavidCarlisle (sorry, that I don't always code for maximum performance...)

@Skillmon yes I just thought it funny coming from you who quibbles about me using an extra \expandafter here and there:-)

@AlexG yes, sorry I should have mentioned this, I know that the "simple" things don't work, I tried them some time ago. I wonder if you saw some "new" trick. We can't be the only one wanting to set this in the postscript and not per command line (and wanting to set version 2.0).
@DavidCarlisle how does it work? Why doesn't latexmk gets confused anymore?

@DavidCarlisle I know, sorry for nagging here and there :)
@DavidCarlisle also, feel free to post the \unskip\space\ignorespace solution :) I promise you'll get a vote from me.

@UlrikeFischer the main thing was that context doesn't seem to pick up the --output-directory flag so the log files were not being written where overleaf expected (not as I thought that the format was too bad) so now I use its --result option in place of --jobname so it writes output.log and output.pdf then just need to move them to the /compile directpry an dit all works, more or less.

@UlrikeFischer Maybe, someone subscribed to gs-devel mailing list could ask there?

10:38 AM
@DavidCarlisle so a variant of "tools not finding their input in output directories"?

@UlrikeFischer output directories should never have been implemented:-)

@DavidCarlisle like spaces in pathes. But people love them. ;-(

@UlrikeFischer spaces between words, surely not a German trend. :)

@PauloCereda This starts with the Vimtex vim package starting from the wrong place, what do people expect:-)

11:00 AM
@DavidCarlisle as promised.

@Skillmon :-)

Welcome, @nbro!

Hello

@Skillmon only fair to reciprocate

@DavidCarlisle ohh, I'm flattered :)
/blush

11:15 AM
@Skillmon of course it's not actually deserved, if that helps your embarrassment. I just didn't want to be seen as unfair:-)
2

@DavidCarlisle you're so considering!

@DavidCarlisle all important packages mentioned in the tug message ;-)

@UlrikeFischer ooh first place in the list

@DavidCarlisle sorted by age?

11:46 AM
Need some help about a design question. Which of the following two would you prefer?

@Skillmon Hmmm... At first glance, the first one. But I suppose that in the second the first url belongs to Tobias but Jonathan has no corresponding one, and the ones after are the urls that they have in common. Which seems more intuitive but doesn't look as good. :p

@Plergux exactly my problem... :)

@Skillmon I think I would probably rather go with "looks good" :p If people go "ooh!" and then look at the urls properly they'll see it soon enough anyway. :p

@Skillmon none, the names are to much to the right, they don't look centered. If Tobias name where more right, it would look less unbalanced, and then the web address would not stick out so much.

12:08 PM
@UlrikeFischer the gap between the names is centred, it looks unbalanced just because my name is so much longer.

@Skillmon yes I know. Technically it is centered, but not visually. Gaps count much less than black ink.

@UlrikeFischer importance I assumed

@UlrikeFischer agreed. Producing third example (but first, lunch)

12:24 PM
@Skillmon seriously I'd add a URL for you or (better) drop the one for Tobias. How many contact address do you need on a latex package? It will always look unbalanced if you don't use the same address forms.

12:38 PM
@Skillmon you should also remove the /en (or as suggested by David drop the address)

@UlrikeFischer well, unlike me, Tobias has a website that advertises his services as a (TeX)designer. I will not remove that link from one of his packages.

@Skillmon yes, I know. But you can get it from the mail address (or find the mail on the website).
@Skillmon or you could link to ctan.org/author/spratte

1:01 PM
@UlrikeFischer either that or just stack the authors vertically with a \medskip.

1:13 PM
@UlrikeFischer including a link for myself as well (the CTAN one), gives an even more unbalanced look (no matter whether I centre the gap or the block as a whole), I think I'll stick to the vertically stacked variant.

@Skillmon Simply ask your co-author to change his name :) A double name should make the title look much more balanced :)

@samcarter_preparing_for_xmas that seems like the easiest solution indeed. But I don't consider myself a co-author. He's the author, I just do some maintaining. That might change in the future (I'm currently trying to rewrite the package), but that's how I see it for now.
4

@DavidCarlisle I get compile timeouts on it?! Must have been a blip, works now.
@DavidCarlisle btw, there is a clear and justified pushback to supporting ConTeXt officially, e.g. with a Menu option: when you support it, you have to support it.

1:32 PM
@Skillmon I'd prefer this one.

@AlexG +1 @Skillmon

@AlexG yes, that's the one I'll stick with, thanks for the input (also thanks to @Plergux, @yo', @UlrikeFischer, @DavidCarlisle, and @samcarter_preparing_for_xmas)

@Skillmon You're welcome. That's what this place is for, right? :p

@Plergux no, this place was made only to remind @CarLaTeX that there is pineapple pizza. I'm 100% sure that's the purpose of this chat.

@Skillmon lol, no objection from me. The world needs to to know the beauty of pineapple pizza :p

1:58 PM
@Plergux TANSTAPP: There Ain't No Such Thing As Pineapple Pizza.

2:16 PM
@HaraldHanche-Olsen Really? Oh well, guess I'll just have to have Banana Pizza instead :p

@Skillmon Not only, sometimes we talk about ducks

@CarLaTeX yes, those tasty, crispy ducks :)

@Skillmon waiting for @DavidCarlisle saying "dinner"

@CarLaTeX and for @PauloCereda saying "ohh no"

@Skillmon Exactly!

2:50 PM
@CarLaTeX it's before tea time, so it can't be dinner!

Deja vu? Anyone? :p

yesterday, by David Carlisle
3 hours ago, by Harald Hanche-Olsen
must … not … link … to infinitely deeply nested “deja vu” comment …

@Skillmon XD yeah, we all know how that ended :p

22 hours ago, by Plergux
@DavidCarlisle Oh, I'll go all the way down to the Bottom and stay there quite happily :p

@HaraldHanche-Olsen :D That is one of my two biggest qualities; wallowing in puerile filth and taking nothing seriously. :p
@HaraldHanche-Olsen Which does not mean that I'm not serious about my work, only that it can always be made more funny :p

3:45 PM

@FaheemMitha TeX Live distribution.

@AlanMunn @Skillmon Not if your TeX distribution comes courtesy of Debian.

@FaheemMitha You've been around here long enough to know that you should install a vanilla TeX Live instead. :)
2

@AlanMunn I'm a slow learner.

3:50 PM
@FaheemMitha Although is that really true? I think one of the TeX Live maintainers (Norbert Preining) also maintains the Debian packages.

I'm wondering if fontspec has an obsolete warning, or if something else is going on.

To get hyphenation with \ttfamily with _lualatex_ I declare HyphenChar with fontspec as suggested by egreg's answer here: https://tex.stackexchange.com/questions/290731/

When I do, hyphenation works (yay), but I get this warning:
Package fontspec Warning: Ignored XeTeX-only feature: 'HyphenChar'.

Could the warning be obsolete?

\documentclass{article}
\usepackage{fontspec}
\setmonofont{Latin Modern Mono}[HyphenChar={-}] % need for hyphenation

@AlanMunn Well, I don't actually remember what a biber binary looks like. But the Debian package requires a newer version of Perl to build than is available on Debian.
It's not the first time. I've been through this before.

@FaheemMitha Then I repeat my previous message: install a vanilla TeX Live and don't bother with the Debian packages. It's true the Debian base list doesn't seem to contain a biber binary. I have no idea why.

@AlanMunn Well, biber is contained in a separate package.

@dedded the answer is for xelatex.

3:57 PM
@FaheemMitha That doesn't answer why that's the case though. It's not clear why it's not included in the base binary package.

@UlrikeFischer Understood, but it works with lualatex (I just get the warning).

@dedded well as the message says it is ignored, so probably it works only without it.

@UlrikeFischer No, I tested that. The fact that it works made me wonder if perhaps fontspec had been enhanced, but accidentally left the warning in place.

@dedded It's not obsolete. If you take the same document with \setmonofont{Latin Modern Mono}[HyphenChar={-}] replaced by just \setmonofont{Latin Modern Mono} because fontspec by default enables hyphenation of the monospace font in LuaLaTeX if loaded using \setmonofont.

@dedded hm, yes from the code it looks like this. You could open an issue.

4:04 PM
@MarcelKrüger Like I said, I tested that (and just now re-tested). \setmonofont{Latin Modern Mono} without HyphenChar does not enable hyphenation.
@UlrikeFischer On fontspec, or lualatex?

@dedded It does for me. Which version of fontspec do you have?

The log file reports this (I've done a tlmgr update within the last two weeks):

Package: fontspec 2020/02/21 v2.7i Font selection for XeLaTeX and LuaLaTeX
Lua module: fontspec 2020/02/21 2.7i Font selection for XeLaTeX and LuaLaTeX

Also in log file:
Package: fontenc 2020/08/10 v2.0s Standard LaTeX package

@Plergux -- Ah! The sheep farmer's answer to prairie oysters. Sounds like quite a spread.

Ugh, just realized that last says 'fontenc' not 'fontspec', but I can't re-edit the comment.

@MarcelKrüger the option works, it only produce lots of warnings.
@MarcelKrüger what is the official method to separate the comma here from the word and so change direction? tex.stackexchange.com/q/573746/2388

4:13 PM
@dedded Sorry, had a wrong fontspec file in my working directory.

@barbarabeeton lol, exactly :) It is. Then you of course add sweet rye bread and dried fish with tons of butter and flush it all down with brennivín and you've got a winner. :p

@dedded Anyway the warning is still mostly right. In LuaLaTeX, it does not actually change the HyphenChar, it can only turn hyphenation on or off. So basically =None works (and does not produce a warning), while everything else is ignored except that it cancels an earlier =None (Which is automatically added for \setmonofont).

@MarcelKrüger tss how could that happen? ;-)

@yo' the first one is very slow (font cache) and can timeout but running it again completes the cache
@yo' yes I can understand that I'm sure you end up answering as many "latex" questions as you do "overleaf setup" ones
@FaheemMitha that doesn't sound right, are you sure about that?

@DavidCarlisle I'd have to unpack a Debian biber binary to be sure.
Is /usr/share/texlive/texmf-dist/tex/latex/koma-script/scrpage2.sty the correct location for scrpage2.sty?
Because my LuaTeX is not finding it.

4:28 PM
@FaheemMitha yes assuming your texlive is rooted at /usr/share/texlive/texmf-dist

@DavidCarlisle Yes, it looks like the correct location.

@FaheemMitha did you copy it there by hand or is that just your texlive install?

@FaheemMitha if you compile a small document with \documentclass{article} what does the log say about the path of article.cls

So, ideas what went wrong? The Debian packaging system should update the indices or hashes or whatever, but I could run it again.
@DavidCarlisle Part of my upgraded TeX Live install.
@UlrikeFischer With LuaTeX?

@FaheemMitha yes (not that it matters)

4:31 PM
@FaheemMitha run mktexlsr again just to be sure, or maybe you wrote \usepackage{scrpage3} by mistake. How can we telll as (not for the first time) you are showing no code.

@DavidCarlisle It's old, prexisting code. I was running a test suite.

@FaheemMitha if tex is behaving differently its input is different, so if you don't show your input or the log of what tex did, we can't really help. But you know that.

@UlrikeFischer /usr/share/texlive/texmf-dist/tex/latex/base/article.cls
@DavidCarlisle I think debugging will have to wait till tomorrow.
It's probably bed time for me now. I'm trying to stick to a schedule. I hear that's good for me.

@FaheemMitha then the path should work.

@UlrikeFischer It it would follow the Unicode standard algorithm you could insert U+200F (Right to Left mark) but if it would follow that algorithm the problem wouldn't exists in the first place. So ... use LuaTeX with a proper bidi implementation?

4:40 PM
@MarcelKrüger why wouldn't it exist in the first place? I thought as it is after the LaTeX, it would get its direction from it and so be printed in LTR too?
@MarcelKrüger but you are right, with luatex + babel the comma is at the other side, but I won't claim that I understand the rules ;-(

@UlrikeFischer Resolving neutrals in the bidi algorithm is a bit more involved. Basically they take the direction of the surrounding characters only if the surrounding characters on both sides are the same(ignoring other non-strongly directional characters like spaces). Otherwise they get the embedding direction. Since there are no explicit embedding characters involved, this is effectively the paragraph direction which happens to be RTL.

@MarcelKrüger that makes sense. I think I will a luatex answer, or do you want to do it?

@UlrikeFischer Feel free to write an answer, I'm busy with other stuff.

@MarcelKrüger ok, will do (I tried already U+200F in the xetex case, but it didn't work).

5:34 PM
@DavidCarlisle ah right!
@DavidCarlisle yeah. We provide (limited) LaTeX support.
@DavidCarlisle btw, any idea if it were possible to time the font caching? We have some idea that it would be useful to exclude the font caching from the timeouts...

6:13 PM
@yo' For which engine/format?

@yo' certainly would be easy for anyone who knew the context runner script to make it give timings
@MarcelKrüger context

I think there is already some timing code there, so it should just be a matter of increasing the log level enough.

@yo' @MarcelKrüger presumably you could populate the lua font cache in the docker image that the first context run doesn't take such a hit. (Rather than letting it be slow but not count to the time limit)

6:31 PM
@DavidCarlisle @yo' That would translate to (since there is no reason not to rebuild luaotfload's cache too while you're at it): mtxrun --script font --reload --force && luaotfload-tool -fu

6:43 PM
@Plergux -- The next time we visit Iceland, we shall ask you ahead of time where are interesting places to eat. On another topic, do you know whether there is any group of (La)TeX users in Iceland? There used to be, a long time ago. I keep hoping that someday there will be a group well enough organized to host a TUG meeting. (Sometime when we can all travel again ...)

6:58 PM
@barbarabeeton I know there are some. But most documentation that I have found in Icelandic is fairly old. Maybe I should just find out :D

7:14 PM
@Plergux -- The web site listing TeX groups around the world still lists one in Iceland, and there is still an intact web page: istex.hafro.is The last date mentioned there is 2014, but maybe there are still a few people who would respond if poked. There are still a couple issues of the newsletter SamanTeXt online, and the topics covered looked like things I knew something about (even if I can't read them). But there's definitely a dearth of hyphenation; I am unfond of \sloppy.
I hope the hyphenation gap has been filled. That's something you surely know about by now.

@barbarabeeton I'm using ctan.org/pkg/icehyph which seems to be working fine :)

@Plergux -- Glad to hear it!

7:41 PM
Does anybody know how to make multicols* work inside another environment? Sort of opposite of \relax :p

@Plergux I don't think it's possible. What are you trying to do? There may be better ways than multicol.

@Plergux which environment?

@AlanMunn I'm trying to create wordlists, @UlrikeFischer, with a custom float and tcolorbox. I have several word lists, e.g. examples of compounds, and if the number doesn't divide equally into the columns the fewer words in the last column stretch down to the end, which I don't want. Like, if I have eleven compounds and I tell multicols to do three columns there are four and four equally spaced and then three that have more space between them and I want them to be equal.

@Plergux I think you want \raggedcolumns

@Plergux You may want to try the tasks package. Or a table might be a better choice. It's not hard to auto-number a table, if that's your concern.

7:57 PM
@PhelypeOleinik ooh! I'll take a look at that.

@PhelypeOleinik I don't think this is sufficent to line things up properly.

@Plergux but if multicols work, why should multicols* not work?

@UlrikeFischer Try, e.g. inside (not outside) an enumerate environment.

@AlanMunn If I understood @Plergux's problem right, \raggedcolumns (with non-starred multicols) will keep the alignment in columns with fewer items (as long as all items have the same height, of course). But a dedicated package might be better indeed

@AlanMunn Tasks looks promising. And yes, I found multicols to be easiest because some of the lists have five words and others have fifty. :|

8:01 PM
@PhelypeOleinik Yes, in principle, but I'm assuming they're using some sort of enumeration, and then things get tricky. But maybe @Plergux should show us some minimal code... :)

@AlanMunn on it! :p
@AlanMunn how do I type code in here again?

@Plergux Just paste it (not as a reply) and then ctrl-k

\documentclass{book}
\usepackage[T1]{fontenc}
\usepackage{tcolorbox}
\usepackage{multicol}
\usepackage{float}

\floatstyle{plain}
\newfloat{wordlist}{tbhp}{lol}[chapter]
\floatname{wordlist}{Orðalisti}

\makeatletter
\newenvironment{bigwordlist}
{
\refstepcounter{wordlist}
\renewcommand{\caption}[2][\relax]{
{\center\textbf{\fname@wordlist~\thewordlist} ##2\par}
\ifx\relax##1\relax
\else
hmmm... did that come out right?

8:16 PM
@Plergux Yes, that came out right. But honestly I would use a tabular here instead.

@AlanMunn But then I have to calculate how many go in each column :(

@Plergux Out of curiosity, what do the numbers represent? Does it make sense that they're not aligned?

@AlanMunn They're frequencies. I don't know if it makes sense, it's just how it came out when I was thinking how I should represent them. The numbers themselves aren't important as such. Each list is just sort of "This is a list of words that has a frequency in this range".
@AlanMunn Some of them (like frequency=1) don't have any numbers and they number in the dozens. :|

I would do this:
\documentclass{book}
\usepackage[T1]{fontenc}
\usepackage{tcolorbox}
\usepackage{multicol}
\usepackage{float}
\usepackage{array}
\newcolumntype{B}{>{\bfseries}l}

\floatstyle{plain}
\newfloat{wordlist}{tbhp}{lol}[chapter]
\floatname{wordlist}{Orðalisti}

\makeatletter
\newenvironment{bigwordlist}
{
\refstepcounter{wordlist}
\renewcommand{\caption}[2][\relax]{
{\center\textbf{\fname@wordlist~\thewordlist} ##2\par}
\ifx\relax##1\relax
\addcontentsline{lol}{wordlist}{\protect\numberline{\thewordlist}##2}

@AlanMunn Cool! :D What does \setlength{\extrarowheight}{3pt} and \begin{tabular}{BlBlBl} do?

8:29 PM
@Plergux Using the array package, I defined a new column type to be a left aligned bolded column. The extra row height just increases the spacing to match (roughly) what your list looked like. It just adds some extra space to each row of the table.

@AlanMunn Ah, yes, I see it now. Read everything first, then ask. :p

@Plergux And it's not too hard to convert your existing code to the table with a regex replace.

@AlanMunn Yeah, that's true :) I'll just have to accept that the easy way out is not the best way :p
@AlanMunn Upside is I'm learning something new every day :D And my thesis is so pretty :p

@Plergux Also if you want the frequencies aligned, make the column r instead of l or use the dcolumn package to align on the decimal point (siunitx can also do this, but seems overkill for this kind of data).
@Plergux There's an expression in Portuguese O preguiçoso trabalha dobrado 'the lazy works twice' :)

8:52 PM
@AlanMunn That sounds pretty accurate. :) Though in this case I will argue that it's more that I'll try anything not to have to do any math. :p (i.e. calculating how many words go in each column :p)
@AlanMunn And I'm gonna have to credit all'yall in my thesis. :p

2 hours later…
10:41 PM
@MarcelKrüger (Two power outages and 6 hours of shoveling snow later...) OK, sounds like I shouldn't file an issue then. Thanks

11:00 PM
@MarcelKrüger I was just doing something with the noto color emoji font \newfontfamily\notoemoji{Noto Color Emoji}[Renderer=HarfBuzz] and noticed that if one use a not existing glyph one get a message that it is missing in nullfont: Missing character: There is no a (U+0061) in font nullfont!. I guess this is because it it a special font?
@Rmano ^^^^ we did it!

11:33 PM
@UlrikeFischer Yes, in order to get the right fontname there we would have to allow the glyph node to reach the backend, but then luatex would try to embed the font file and break when it realizes that no normal glyph tables are present. Therefore we switch to nullfont which will be ignored but still triggers the warning.

@MarcelKrüger I guessed so. It is not really a problem, but it confused me at the begin, as I had tikzpictures everywhere and I thought I landed in one of them.

@UlrikeFischer I'll look into getting better warnings here. One option is the glyph_not_found callback but it's complicated because that one is rather limited. We could replace them with whatsits but that would have linebreaking and hyphenation effects. So the best way might be to dynamically define such glyphs as virtual glyphs emitting warnings.

@MarcelKrüger it would be nice, but I wouldn't put too much time in it . The number of users falling over this is probably rather small.