@DavidCarlisle Glad to see myself proved wrong on this one.

@JosephWright Pretty much support for being able to do ICU case folding. Also primitive-level support for the text language, so that if a section header is uppercased, for example, the text language will be identifiable at that point.

One of the important things is that the document language will look like LaTeX but the programming language will not. I'm some ways away from even committing to what the programming language will be, but I'm expecting to steal some other language wholesale, the leading contender at the moment being a dialect of JavaScript.

@DavidCarlisle You may want to properly duplicate-close them or post a new answer?

@DonHosek -- You may want to see what you can find out about the failed NTS attempt. If I remember correctly, it was using JavaScript, although it may have been Java. The main complaint that I remember was that it was much too slow. A lot of people these days seem to care about speed almost more than they care about the final result.

@barbarabeeton NTS was Java. At the time, the start-up time for Java was not so great and byte code interpretation was subpar. Contemporary Java is competitive speed-wise with native code, although there's still a smallish start-up cost which makes it more amenable to long-running processes like web services than run multiple times applications like TeX, although things have improved since the days of NTS.

I'm writing in Rust which gives me the speed of writing in C/C++ with better guarantees about memory safety and multi-threaded code.

@barbarabeeton It occurs to me that my language was imprecise talking about JavaScript as the programming language. It's not meant to fill the roll of web/web2c in TeX but the role of macros and expl3 in LaTeX.

@DonHosek -- I'm not a really fluent programmer, but I understand the flow of TeX well enough that the only real opportunity I can see for multi-threaded code is, say, compiling graphics or things of that sort that are not directly in-line. I've never actually heard of Rust, so I have to take your word for its being a good choice. Thanks for the clarification on Java/JavaScript.

@PauloCereda -- I've just tried to help someone who's using the abntex2 package. (tex.stackexchange.com/q/615273 ) You're correct that the documentation isn't very good. Please look at my last comment and tell me whether or not I got it right.

@samcarter -- I just got a popup of the TUG'21 community promotion ad. I have posted a whimper at meta.stackexchange.com/a/369797 but honestly don't expect much. The detail of the 2020 and 2021 ads has been removed from meta, so it can't be cloned again for next year. I guess we should just consider the whole effort dead. I am disgusted.

Is this good typography, and is \mathrel{\phantom=} a good way to get in TeX?

  1 + 2 + 3
= 4 + 5
= 6 + 7

It has probably been asked before but I don't know what it's called, and it definitely isn't covered in amsmath doc

@DonHosek Sure. That I guess links to the fact that case changing can come in both areas. Users want to uppercase for design reasons, and programmers want to manipulate 'strings'

@DonHosek The Team have talked about the fact that for the typesetting case, one might want to treat case as a font property rather than a text change - but that's tricky as it can affect line length (so has to be handled before hyphenation)

@DavidCarlisle (regarding tex.stackexchange.com/questions/615213/…) yes, just wanted to point out that the answer is not "perfect" (because that is obviously not possible) :)

@barbarabeeton Yes, I think we can agree that expectations for any typesetting software is high, and extreme flexibility is a must. And you didn't even mention performance issues, which are extremely critical in practice. If it is too slow, for some value of "too", people will simply not use it. I'm guessing your comment comes from direct experience.

@barbarabeeton Yes, I can imagine that they would treat it like they would a word processor.

@DonHosek I thought that both LuaTeX and XeTeX now do everything in UTF-8.

@user202729 you can use an alignment rather than phantom but I see no reason to align the left hand side with the right hand side that just confuses. If the column with is too narrow OK to break theer but I'd set the first line to the left

@Skillmon :-)

@FaheemMitha UTF-8 isn't the issue the point here is that in most locales the uppercase of i is I but in Turkish it is the eminently reasonable, İ

@DonHosek not a bad choice but the community would probably prefer Lua, given that between luatex and SILE there is some Lua knowledge building up. Although javascript might lead the way to in the browser versions at some point down the road, which might be interesting:-)

@FaheemMitha Take a look at SpecialCasing.txt (available in your TeX system, so find it using kpsewhich)

@JosephWright I see it. Those are presumably hexadecimal codes.

@FaheemMitha Yes

@DavidCarlisle That also works with pdflatex here. Is expl3 dependent in any way on LuaTex?

@FaheemMitha No

@JosephWright OK. Good.

@DavidCarlisle, @UlrikeFischer, @egreg Thoughts on github.com/josephwright/siunitx/issues/547? Basically what's happening is in v2 I always use \textmu but for v3 with Unicode engines I expect a 'real' µ to be available

@JosephWright well imho using T1 encoding with the unicode engines is an user error. But probably newtxtext should be adapted to recognize the engines and at least warn or error.

@UlrikeFischer Yes, I'm thinking this is a 'it's not my fault' case, but it's borderline (as things used to work). I'm tempted to document that siunitx assumes TU (for XeTeX/LuaTeX) and if that's not the case then users need to set stuff up manually.

@UlrikeFischer I had a look at newtxtext: it's quite complex

@UlrikeFischer I also have github.com/josephwright/siunitx/issues/545: again I'm thinking this is on the not-my-fault border

@JosephWright especially with xetex usig (O)T1 with xelatex and having all the hyphenation wrong seems like user error. It would be nice though if the font packages did the reasonable thing like lmodern of using a suitable opentype version with luatex and xetex.

@JosephWright yes, adding support for lualatex would be perhaps too much work, but a warning would be good. Did you already contacted the author?

@UlrikeFischer Not yet: I'll do it later on

@DavidCarlisle Indeed

@JosephWright yes, they shouldn't expect you to support such encoding mixes.

@JosephWright you could perhaps add a test for the encoding at begin document and issue a warning if it is not TU ...

@UlrikeFischer I think I can avoid the hard error, now I think about it: probably I'll adjust

@UlrikeFischer because obviously using an 8bit font for mu is wrong in xetex given that the default encodings for xelatex math are OT1 and OML

@JosephWright one can always avoid that but what are the costs? Will it also work if someone switch to T1 in the middle of the document?

@DavidCarlisle ;-). But that one was with polish and luainputenc and T1.

@UlrikeFischer we could add \PackageWarning{luainputenc}{are you sure you want to be using this package} to the package.

@DavidCarlisle We could add this to firstaid and use a package hook for a number of packages ....

@DavidCarlisle One day we may need to fix that

@JosephWright need opentype math to be a bit more stable but then yes

@UlrikeFischer Er, that won't work so well - I could I guess go back to the older approach, but I'm really trying to write code that looks forward not back

@DavidCarlisle Yes, I meant mainly that with tagging, etc., this is something that perhaps we should be at least having on the agenda

@DavidCarlisle I wonder if we might also look again at OT1 in pdfTeX ...

@JosephWright I would require TU encoding for text.

@UlrikeFischer Fair enough: I will think about adding an error

@DavidCarlisle You want to take today's LaTeX2e/LaTeX3 issue? Or should I?

@JosephWright defaullt to T1 and set the pdf page size in any document that has any recognisably new markup at the start....

@JosephWright the expl3 rollback one? for you or Frank I think

@DavidCarlisle OK :)

@DavidCarlisle Yes, something like that as part of the tagging project

@DavidCarlisle I could do that in the pdfmanagement, but imho we are still missing backend functions for the page size. @JosephWright

@JosephWright Well, using newtxtext with XeLaTeX is a no-go.

@egreg yes but the question really is, Is it reasonable for end users to know that given that superficially similar packages like lmodern are fine.

@egreg @JosephWright @UlrikeFischer would be interested in thoughts on my answer and the subsequent comment discussion here tex.stackexchange.com/a/615262/1090

@DavidCarlisle @JosephWright lmodern is a particular case, because it doesn't set the encoding to T1, so TU is chosen because the engine is XeLaTeX. What strangely happens is that the mu is taken from the T1 font rather than TS1

@barbarabeeton arrrg! And the flag to please delete the old add on meta is still pending -- since 2nd of September :(

@egreg yes that, plus the fact that the classic nfss lm familiy has pre-existing fd files for TU that set up the right font.

@DavidCarlisle not quite sure if the code is executed always in the place where he wants it, but I think something like this should reproduce his use case

\documentclass{article}
\let\testenv\document
\let\endtestenv\enddocument
\AddToHook{env/testenv/begin}{\def\foorbar{1}}
\AddToHook{env/testenv/end}{The End}

\let\testenvB\document
\let\endtestenvB\enddocument
\AddToHook{env/testenvB/begin}{\def\foorbar{2}}
\AddToHook{env/testenvB/end}{The End B}

\begin{testenvB}
Test\foorbar Test
\end{testenvB}

@UlrikeFischer thanks

@barbarabeeton will take a look, thanks. :)

@PauloCereda sudo dnf system-upgrade download --releasever=35 --allowerasing ^^

@PauloCereda Linux rayzentex 5.14.1-300.fc35.x86_64 #1 SMP Fri Sep 3 16:27:33 UTC 2021 x86_64 x86_64 x86_64 GNU/Linux

Nada como un pato actualizado :D

I wonder, is it \TeX~Live or \TeX\,Live?

@yo' texdoc texlive uses a full space.

@DavidCarlisle oh such a thing exists! It seems not to be linked from ctan.org/pkg/texlive

thanks!

@yo' all 44 pages of it:-) But it needs to be hidden to protect users from documentation

@DavidCarlisle right right

I'm just rewriting some of our internal documentation, so bad things are happening at the moment!

@yo' actually it's more of an "installers guide" than a users guide so at least in classical university installations not that relevant to most users

zoom has the worst kerning for names of participants ...

@samcarter say, beautiful!

I bet even ransom letters with cut out letters from newspapers will do a better job :)

@yo' yes, much better :)

@samcarter 𝚆𝘩𝗮𝕥 𝒊𝕤 𝕨𝕣𝙤𝘯ℊ 𝖜𝙞𝗍𝘩 𝘁ℎ𝚎 𝖙ℯ𝖝𝒕?

@DavidCarlisle nothing wrong, I just did not know that the design service you use for the typewriter package also works for zoom :)

@UlrikeFischer That goes a long way. Thanks a lot. My immediate problem is solved. However, any font family command issued in the env/.../begin hook is overruled. I managed to solve this in my particular case by moving that to an \AtBeginDocument call, but this only works because all the environments in my particular class use the same font.

@George you can add hook commands:

\AddToHook{env/testenvB/begin}{\AddToHook{begindocument}{\sffamily}\def\foorbar{2}}

@DavidCarlisle I thought kerning was a lower case thing.

@FaheemMitha those from cities on the river VLTAVA know that this is not true :)

I've never really understood why kerning is considered important, though I gather it's an aesthetics issue.

@FaheemMitha it's legibility issue too.

@yo' Oh?

@FaheemMitha badly spaced letters slow your reading, obviously

@yo' Yes, they can be distracting.

@FaheemMitha and that's the point ;-)

@yo' Still aesthetics, though.

Does harfbuzz handle kerning now?

@FaheemMitha if you say so. I'm happy to know that proper design can save lives -- maybe not in kerning, but certainly e.g. in traffic sign design and placement. The same goes with car front desk ergonomics, or even plane cockpits

@yo' I'm a fan of aesthetics. And I didn't say it wasn't useful.

@FaheemMitha ok, we're in agreement then :-)

Or valuable, perhaps.

@FaheemMitha not just aesthetics, it can change the meaning if you misread m for rn

@samcarter Is that an issue? I thought was an issue only for handwriting.

@FaheemMitha boredpanda.com/funny-letter-spacing-fails

Many, many people should not be writing stuff by hand. And if they must, they should print it. Carefully. And not just doctors.

@samcarter Some of those are pure spacing issues, though.

@FaheemMitha yes ?

I thought kerning was about joining letters together.

@FaheemMitha you mean ligatures?

@samcarter Oh, apparently I didn't know what kerning meant.

@samcarter Yes, like ff, for example. Or ti.

@FaheemMitha kerning means spacing hence Tex \kern 5pt

@DavidCarlisle Yes, I realised that just now. I thought it was what ligatures is called.

@FaheemMitha OK, but even ligatures are not just for aesthetics. They influence readability of texts and even actively carry information about word boundaries (important for languages which save up their white spaces...)

@PabloGonzálezL WAIT WHAT

@PabloGonzálezL ah beta. :) How do you like it?

@PauloCereda Great, in general there are not many changes at visual level and it works perfectly :D

@PauloCereda In general, beta versions are very well polished and there are very few changes with respect to the final version.

What's great is the upgrade system (which I know you don't use :)

@PabloGonzálezL I will probably do this time :)

@PauloCereda It only took half an hour on my note :D

@PabloGonzálezL I finally managed to get a great setup, so I will bite the bullet this time for F35. :)

@PabloGonzálezL ooh

@PabloGonzálezL what's the ETA for final release?

@PauloCereda About 3 weeks to go

But that is relative as always :D

@PabloGonzálezL cool, I was worried that I missed the announcement. :)

"19th October 2021, with a delayed date for 26th October 2021..."

@PabloGonzálezL which we know it will happen. :)

@PabloGonzálezL which noteworthy changes to expect in this version?

Gnome :D and many lib update

@PabloGonzálezL ooh :)

Python and other (remove old stuff)

@PabloGonzálezL ewww pythons and other snakes :)

@PauloCereda I think libadwaita its most imortant

@PabloGonzálezL quack :)

@PauloCereda What I can say is that the startup goes ultra hyper fast.

@PabloGonzálezL ...and this is something that makes me think about abandoning Ubuntu... in my home PC is now painfully slow to start. Partly is due to onedrive daemon, but still...

@Rmano :)

@PabloGonzálezL ooh :)

@Rmano Once you try Fedora, you won't go back :)

@PauloCereda I'm quite sure, although I fear a loss of time (that I don't have) to adapt back from apt to rpm.

Or whatever they have now ;-)

But i will give it a spin for sure

@Rmano fedora mate simple, accurate and fast

@PabloGonzálezL isn't mate a gnome2-similar thing? I do like gnome3...

@Rmano dnf :)

@PauloCereda Do Not Fail? A good plan ;-)

@Rmano quite :)

@Rmano Then the default installation of Gnome 41 will suffice :D

@Rmano you could alias apt to dnf -- problem solved :)

dnf = @david Not Fault :D

@PabloGonzálezL :)

@samcarter I believe you. I know little or nothing about typography. I know what I like, though.

@PauloCereda oh so this machine at 5.12.15-300.fc34.x86_64 isn't too far behind?

@DavidCarlisle nope :)

@PauloCereda you wouldn't like my window manager though:-)

@DavidCarlisle you sound like you use XFCE :)

@PauloCereda windows 10

@DavidCarlisle oh no

@PauloCereda How is your new computer holding up?

@FaheemMitha it's being a fantastic experience, the machine is excellent and the performance is beyond my best expectations. :)

@PauloCereda As long as it doesn't become sentient, that's all good.

I am not used to a computer turning on and off in less than 5 seconds. :)

@PauloCereda obviously doesn't run tikz on startup

@DonHosek I was wondering what you thought of SILE, if anything. And whether you had considered contributing to it, instead of starting a new project.

It seems to be one of the few (possibly the only) free software typesetting project that is currently showing any activity. Aside from TeX, of course.

Looks like basic math typesetting just got added.

@FaheemMitha Well, there's github.com/speedata/publisher as well (partially based on LuaTeX, iirc). So definitely not the only other free option ;)

github.com/sile-typesetter/sile-typesetter.github.io/issues/19

@JosephWright I think it definitely needs to be a font property. The edge cases can drive you nuts, like ᾀ will capitalize differently depending on whether it's a title case (ᾈ or Ἀι depending on the font and preferences) or in an all caps context (ΑΙ or less frequently ἈΙ)

@TeXnician Thanks for the pointer.

@TeXnician Yes, that one seems to be somewhat active too, but more of a 1 person project. Though SILE is still relatively small, it still has quite a few people working on it.

@DonHosek I implemented the expl3 case changer, so I do know about the the edge cases ;)

@TeXnician Very different to a LaTeX-like approach, though

@DavidCarlisle of course, run pstricks with Lua on startup, that's the secret

I've never seen a baby typesetting project in the wild before, so I'm a little curious.

TeX has seen remarkably little competition on the free software front for 40 years or so now.

@PauloCereda Is it a "yes" to my email? ;-)

@FaheemMitha nroff, lout, fop, texmacs, patoline, .... it depends how you define "competition"

@StefanKottwitz ooh

@StefanKottwitz I will reply, but I am inclined to say yes. :)

@FaheemMitha The glib answer to that is that I'm too egotistical to contribute to something I'm not running. That's probably the serious answer too. Although I did look at it, and I think that I disagree with a lot of how they're approaching things. I have no expectations that I'll do anything other than end up on the dustheap of the internet, and if I'm wrong, I get to be pleasantly surprised.

@FaheemMitha Knuth wrote TeX by himself. :)

@PauloCereda I suspect he had some help, but I take your point. But more generally, projects have a better chance of survival if they have multiple people working actively on them. Otherwise one has a single point of failure.

@DavidCarlisle I still stay with "remarkably little competition".

@DonHosek In general it's better to contribute to an existing project than start a new one. Statistically speaking it's more likely your work will not wasted. But you already know that.

@PauloCereda can you give me the output of systemd-analyze?

@PauloCereda Not exactly Luis Trabb-Pardo, Frank Liang and Michael Plass all had contributions to the code. (As an aside, I have a copy of Literate Programming next to my desk that I ordered earlier this summer and just discovered my name in the index.)

@DonHosek I've always assumed Knuth's grad students helped. Though I don't actually know if that is true or not. I know he originally gave the project (in its early stages) to some grad students to work on, and it didn't go so well. But I don't know what happened afterwards.

@FaheemMitha just because somebody wins doesn't mean that there wasn't a competition

@PabloGonzálezL hold on

@DavidCarlisle Over the period I've used TeX (from around the mid-1990s) I've never got the sense there was any alternative to it. And I've looked, occasionally, though I've historically not had much interest in typesetting. But I like to experiment.

@FaheemMitha well in practice the main challenger has been Word and by numbers that's winning by a few orders of magnitute. By choosing the parameters of the competition you can always arrange that you are the winner if you wish.

@PabloGonzálezL vvv

$ systemd-analyze
Startup finished in 6.266s (firmware) + 2.104s (loader) + 3.653s (kernel) + 1.233s (initrd) + 12.347s (userspace) = 25.604s
graphical.target reached after 12.328s in userspace

@DonHosek Thanks. :) Still, it's Knuth's TeX, not LTP+FL+MP+DEK's TeX. :)

Of course troff/groff for example has been around for a long time. And I think scientific papers have been written in it too. But it never really seemed like an adequate substitute, at least to me.

@DavidCarlisle My interest is strictly in free software. I tried Word around 1998/1999, when UNC saw fit to inflict Windows machines on us grad students, and ran away screaming. It was quite horrible. I found Word entirely unpredictable. TeX was a model of predictability by comparison.

@PauloCereda mmmm

^^

@PabloGonzálezL show off :)

@FaheemMitha I can't use Word at all, I think I have written one document in it, and that was under protest (I have it on this machine but only as a reader really) but as I say if you choose the parameters of the competiion to discount the competitor who is two laps ahead you can declare victory in the race.

@PauloCereda I thought you had a new toy :D

At that time, I didn't really have strong opinions about free vs proprietary software, but I still found working with a black box very uncomfortable. I still can't understand how people can stand it. But I suppose it helps if you have no interest in typesetting per se. You probably end up being more productive.

@PabloGonzálezL you have more money, I don't. :)

Like people who have no particular interest in English, so they don't insanely proof-read their emails, like some of us do.

@PabloGonzálezL this is the best toy I could afford, not the best toy I could get. :)

@DavidCarlisle I'd be interested to learn what parameters would make anything but TeX the winner in the free typesetting race. :-) Though it's hardly a race, IMO. Mostly it's just TeX strolling along, taking the air.

@FaheemMitha you inserted the main one, "free" already as that excludes all the actual competition Word, in-design, 3b2, ...

@PauloCereda nonono...it was what I saved on airfare and other things during the pandemic in 2020 :D...Rayzen 5 2000 - 16 RAM

@DavidCarlisle OK

*5 400

I've not heard of 3b2.

@FaheemMitha you are supposed to call it APP these days, but noone does

Possibly en.wikipedia.org/wiki/Arbortext_Advanced_Print_Publisher

@FaheemMitha yep

@DavidCarlisle :)

@PabloGonzálezL Mine is a humble i5 :)

@PauloCereda Those are more expensive here

@PabloGonzálezL I cannot complain about this machine. :)

@PauloCereda I didn't have much choice...the pandemic depleted the stock...and my old Dell was no longer up to the task of online classes :(

@PauloCereda I bet if @DavidCarlisle were using fedora, the tab completion would add a name to this line systemd-analyze blame ...

@FaheemMitha A couple of years back an article compared latex to word by letting people typeset a text lots of font/colour changes and measured the time they took. latex did loose in producing horrible looking texts :)

@samcarter Sorry, didn't quite understand your last sentence.

You mean LaTeX did worse?

@samcarter It was a daft 'comparison'

@FaheemMitha yes, the tex users took longer in inserting a lot of special formatting to change fonts and sizes. That's simply not something latex is made for - give latex your text and latex will do the correct formatting. Word users are used to messing with such things.

@samcarter Yes, I can imagine that TeX would do worse there. Or LaTeX. Whatever you want to call it.

@JosephWright agreed :) I don't think they really understood how latex is supposed to work

I think I read a comparison once which had people with no prior knowledge of either Word or TeX try to use both and see which one was easier to learn. I think they were probably students. We can guess how that one went.

@FaheemMitha As long as you can define the criteria, you can make any tool win a competition :D

@samcarter That sounds familiar, somehow. :-)

@samcarter a very wise remark

If you judge a fish by its ability to climb a tree, it will live its whole life believing that it is stupid.

@PauloCereda If you judge a duck by its taste then it will live.. oh perhaps not....

@DavidCarlisle oh no

@PabloGonzálezL hmmmm

% systemd-analyze
Startup finished in 2.872s (kernel) + 51.728s (userspace) = 54.600s
graphical.target reached after 51.711s in userspace

I win:

$ systemd-analyze
bash: systemd-analyze: command not found

@DonHosek we IoT are quite interested in your project, and thanks for adopting Rust. :)

@PabloGonzálezL I am not buying this systemd metric, there is no way my machine is taking that much...

@PabloGonzálezL but it is a very interesting command! :)

Hehehe...it is an excellent tool to see what is loaded at startup...from there I see the list and disable things I will never use (like lvm2)

@DavidCarlisle try ` systemd-analyze blame @UlrikeFischer`

@PabloGonzálezL still doesn't work, she must have completely broken my system

@DavidCarlisle wow I'm so effective?

@DavidCarlisle It is the expected output :D

@PauloCereda After playing with systemctl for a while:

@Rmano Sorry :)