TeX, LaTeX and Friends

12:15 AM

@PeterGrill Yes, the code at the end (below “Full example”) works for me. (At least, no errors; haven't looked at the output.)

Peter Grill

@ShreevatsaR Can you complare the output of \listfiles? I must have some package outof date.

ShreevatsaR

12:34 AM

@PeterGrill It gives an error on MiKTeX and on ShareLaTeX, but works fine with my TeX Live 2017. I have older versions of etoolbox.sty, mathtools.sty,

Peter Grill

@ShreevatsaR I am also using TeXLive2017 and all pacakges are up to date as of today.

David Carlisle

12:48 AM

? ! Undefined control sequence.
\shortintertext ...lbaselines \if_dim:w \if_dim:w

Peter Grill

@DavidCarlisle Yeah that is what I get

But is seemed to work for @ShreevatsaR

David Carlisle

@PeterGrill it's one in the morning so I don't care, blame @JosephWright for etoolbox or @daleif for mathtools

Peter Grill

@DavidCarlisle :-) Good night.

ShreevatsaR

@PeterGrill The newer mathtools seems to have stuff like \MH_if_dim:w instead of \if_dim:w … I imagine you need an older mathtools (or rather upgrade Heiko's answer to newer matthools).

David Carlisle

@ShreevatsaR @PeterGrill ah I recall some conversations about making the mathtools fork of latex3 conventions look less like expl3 as it doesn't actually use expl3

Peter Grill

12:58 AM

@ShreevatsaR Replacing \if_dim with \MH_if_dim does not seem like it is enough. Just leads to other errors.

David Carlisle

@PeterGrill ping here if you sort it out or I'll look tomorrow. good night

Peter Grill

1:24 AM

@DavidCarlisle @ShreevatsaR: Looks like that solution works fine with mathools from TeXLive 2016. Won't work with TeXLive 2017 and changing it is not as simple as replacing \if_dim with \MH_if_dim.

daleif

6:10 AM

@PeterGrill heikos answer definitely does not match latest mathtools, which was released in 2017, in it the language it is written in (precursor to the expl3 syntax) was redone so it does not clash with expl3. Not really my problem when people mess with internals

Peter Grill

@daleif Yep, that is what I concluded as well. Unfortinately I am unable to fix as it is in expl3ish form.

daleif

6:33 AM

@PeterGrill it does not seem Heiko is loading expl3 unless etoolbox is. Then it should be enough to prefix all macros with _ in the name with MH, otherwise, simply search for the rest of the macro name in mhsetup and mathtools.

CarLaTeX

7:02 AM

Dear experts, which one is better? ^^^

Is it clear if I put the percentage sign only in the header?

Harald Hanche-Olsen

@CarLaTeX I am no expert, but I have a slight preference for the percentage signs in each entry, although it makes the table look slightly more cluttered. I wouldn' really complain much about the lack of clarity when they're in the header, though. After all, the header is where you look when trying to understand the numbers.

CarLaTeX

@HaraldHanche-Olsen Thank you, I also prefer the one with the percentage sign in every row because these are not percentages from a calculation...

Peter Grill

7:21 AM

:43605848: I tried to replace the` \if_dim` with \MH_if_dim and then other things broke. I'll let Heiko know.

Martin Scharrer

8:40 AM

@CarLaTeX Usually the unit should be in the header. With percent (which is not really a unit) most people might find it clearer if it is with each entry.

Also, you should add a half space between number and unit (using \,)

Or using siunitx package

CarLaTeX

8:58 AM

@MartinScharrer Thank you! I'll use siunitx :)

David Carlisle

geometry update is on ctan:-)

Martin Scharrer

@DavidCarlisle cool

AlexG

9:14 AM

@MartinScharrer Did you also remove the need for luatex85 pkg with the new standalone version?

Martin Scharrer

@AlexG I added the LuaTeX support patch I got.

AlexG

@MartinScharrer Thanks, this will certainly do.

David Carlisle

@MartinScharrer thanks and I just did the same to geometry, so that's probably all the main packages updated for luatex now.

Paulo Cereda

@AlanMunn sadly no, but I can add one. :)

David Carlisle

10:03 AM

LaTeX の inputenc で UTF-8 が既定になった場合？

2

Joseph Wright

@DavidCarlisle Good stuff

@DavidCarlisle Should I talk about it in the blog?

Paulo Cereda

@DavidCarlisle then the LuaTeX team reveals a brand new non backwards compatible engine in TUG 2018. :)

David Carlisle

@JosephWright a bit of publicity wouldn't hurt, I feel it's bound to break something unexpected... (eg I hadn't thought of enctex, and our test suite wouldn't have picked that up if fmtutil --all hadn't complained)

@PauloCereda with harfbuzz, hopefully

Joseph Wright

@PauloCereda Probably

Paulo Cereda

@DavidCarlisle That would be a welcoming headache. :)

David Carlisle

10:08 AM

@PauloCereda although you should not do a major update until you have finished your thesis so texlive 2018 should not be a concern for you

Paulo Cereda

@DavidCarlisle oh no

egreg

10:34 AM

@DavidCarlisle What about Latin-1 encoded documents that only rely on fontenc? I know they're wrong, but I think there are several of them around.

Joseph Wright

@egreg We discussed that

@egreg They'll have to be updated

egreg

@JosephWright Hence we have to be very clear about it and advertise it in big letters. ;-)

David Carlisle

@egreg that's one reason why I improved the detection of bad utf8 sequences to get a better error message for them

@egreg although the test file I added is:

\documentclass{article}
\usepackage[T1]{fontenc}


%\usepackage[cp1252]{inputenc}

\begin{document}

a\"O ``x'' \ss

aÖ “x” ß

\end{document}

@egreg if that was saved in cp1252 then previously you got no error, now you do, but since the old output was nonsense I think we can argue that this is a feature (I know some examples were not nonsense but...)

Ulrike Fischer

10:51 AM

@DavidCarlisle You make it more friendly by using "Grüße" ;-). But does the discussion means that you are planning to add the equivalent of \usepackage[utf8]{inputenc} to the format?

David Carlisle

@UlrikeFischer actually not equivalent, exactly that.

Harald Hanche-Olsen

Out of curiosity, are you talking about LaTeX 3 here, or a future update to 2e?

David Carlisle

@HaraldHanche-Olsen 2e, next week

Harald Hanche-Olsen

Oi. That's going to break documents. But not mine. 8-)

David Carlisle

@HaraldHanche-Olsen why?

@UlrikeFischer:

github.com/latex3/latex2e/commit/…

Harald Hanche-Olsen

11:00 AM

@DavidCarlisle Those that don't use \inputenc? I am pretty sure I have seen them in the wild. Never understood why they work.

@DavidCarlisle Or were you asking why mine won't break? ;-)

David Carlisle

@HaraldHanche-Olsen they work because (in the first weeks of 2e testing) we backed out the error check that stopped them working, but it only works for that subset of latin1 that is the same as T1 so or example it doesn't work for German ß

@HaraldHanche-Olsen such documents will get an error but they can add \usepackage[latin1]{inputenc} or \RequirePackage[2018-01-01]{latexrelease} or they could save the file in UTF-8. Breaking changes are not something we do often but it is that or still be using 8bit encoding default (and not even a standard ISO encoding) for the rest of time.

Harald Hanche-Olsen

@DavidCarlisle I think it's a good change. And it's so easy to fix broken documents, but still I think it will generate waves. The best fix for most users is probably to change the file encoding to UTF-8. Unfortunately, that is not easy for many users.

@DavidCarlisle I think TeXshop, for example, does not let you change the encoding of an existing file. The workaround is to create a new file with the desired encoding, then copy and paste from the old to the new file.

@DavidCarlisle If only everybody was using emacs!

egreg

153

Q: fontenc vs inputenc

I just don't seem to get it. So I come from the german speaking part of Switzerland and naturally I need ä ü ö a lot when I write german texts. So in LaTeX I have the option to go via "a "u "o which isn't really comfortable, or I can use an additional package. So now the confusion starts. I can...

Ulrike Fischer

@DavidCarlisle This means that they will much more active chars from the start on. What should a document do that want to revert this and reset all catcodes to the "normal" state?

Harald Hanche-Olsen

@egreg That's the first time I have seen the advice to load fontenc before inputenc. I must not have been paying attention. Why should it be done that way?

David Carlisle

11:14 AM

@HaraldHanche-Olsen that's partly why I'm discussing it in public here (and opened a public github issue for the same)

@UlrikeFischer currently only using latexrelease to roll back the change. However i did consider adding a new \uspepackage[raw]{inputenc} or some such option with an encoding def file that sets all the 8bit characters to catcode 12, would be easy to add if there is a plausible use case.

@HaraldHanche-Olsen only emacs users really matter

Harald Hanche-Olsen

@DavidCarlisle I suppose it won't bite until TeX Live 2018 is released. (Those who run the beta version will know how to cope.) So you just need to plan your summer holidays to start the day TL2018 is out.

But very often I come across colleagues' machines having TL installations as much as four years old. They just don't feel like changing something that works for them.

Ulrike Fischer

@DavidCarlisle I don't know about a concrete use case but as this is the major change I was considering if there are files which more or less expect the chars to be non-active. E.g. would language patterns break if the chars were already active when they are loaded?

David Carlisle

@UlrikeFischer the chars are active in the patterns anyway as the utf8 patterns use their own decoder, but inputenc is loaded in ltfinal as almost the last thing, after hyphenation, so that should be OK. (at least my format says it's loaded 84 language patterns without complaint)

I suppose I should google to see if there are any answers saying that some package should be loaded after inputenc, as such packages may have problems in the new setup, I'll see what I can find....

Ulrike Fischer

@DavidCarlisle I guessed that you load it late and after the patterns ;-). But it would be interesting to know if the patterns would break if it were before.

Christian Hupfer

@egreg If there are wrong usages of code or typographical doubtful constructs, it's perhaps time to let such document breaks nowadays and force the users that commited this typographical crime to update their documents

David Carlisle

11:29 AM

@UlrikeFischer yes they would as inputenc typically makes characters expand to whatever is needed eg some funky \ooalign constrict to make a comma-under accent and that doen't work too well in patterns

@ChristianHupfer you wouldn't say that if you were maintaining arXiv and had 20 years worth of legacy documents and you had no idea what encoding they are in and may or may not have declared

Ulrike Fischer

@DavidCarlisle on should also consider the packages which changes their behaviour depending on the encoding, e.g. biblatex. They will perhaps get questions from their users.

Christian Hupfer

@DavidCarlisle I would say: Bad luck ;-)

David Carlisle

@UlrikeFischer they should only be affected on documents using undeclared 8bit encoding and really there should be very few of them.

Christian Hupfer

Another palindrome, but @DavidCarlisle will complain about the '+', of course ;-)

Ulrike Fischer

@DavidCarlisle Yes, I doubt that there will be large problems but @JosephWright could ping moewe so that he knows what's going on in case a question pops up.

David Carlisle

11:43 AM

Jun 16 '17 at 12:57, by David Carlisle

@UlrikeFischer first rule of latex maintenance: any change breaks something:-)

percusse

TikZ pie chart jokes #432523

10

Ulrike Fischer

12:06 PM

@DavidCarlisle ;-). I think it is a good change.

egreg

12:27 PM

@HaraldHanche-Olsen It's not really necessary nowadays, but some time ago I got problems when Cyrillic was involved and I found out that fontenc-inputenc-babel was the ordering that solved them.

CarLaTeX

12:42 PM

@percusse LOL

@ChristianHupfer @DavidCarlisle complains about either commas or pluses, it's difficult to avoid his criticisms :):):)

David Carlisle

1:13 PM

@egreg just got a report that it breaks mltex so (like enctex) i need to guard against that...

Transcript for

TeX, LaTeX and Friends

$TeX - LaTeX$ TeX, LaTeX and Friends