@marmot ooh

Stupid question: what is the difference between these to babel shorthands: {"-}{\nobreak-\bbl@allowhyphens} and {"=}{\nobreak-\hskip\z@skip}, I usually use "-, but "= is faster to get in Emacs.

Guys, important question: does your Linux distribution ship pdftk?

I know Fedora stopped two years ago, IIRC.

@PauloCereda at least for Ubuntu 16.04 it is still in the repo, it is not shipped by default.

@daleif Thanks! That's good to know. arara has a pdftk rule and I was curious if I should keep it.

@PauloCereda well, it is not araras responsibility to make sure external tools exists. If the user has access to it, they can use the rule.

@daleif good and fair point, thanks!

@daleif since the command is defined as \def\bbl@allowhyphens{\ifvmode\else\nobreak\hskip\z@skip\fi} and you can not be in vmode after \nobreak- there appears to be no difference at all.

hello

someone help me to write this correctly

@PolineSandra place the &'s in front of the arrows instead

@DavidCarlisle thanks, is it still Johannes who maintains say babel-danish? Because the PDF generated does not mention "= even though it is defined in danish.ldf

@daleif can i obtain x in the middl of B'(0,r)

@daleif @DavidCarlisle In ngerman "- is defined as \declare@shorthand{ngerman}{"-}{\nobreak\-\bbl@allowhyphens} (remark the \- instead of simply a -.). Imho the definition in danish is a typo/bug.

@daleif or perhaps not a bug as it is described this way in the documentation. But it is certainly a way to confuse everyone ;-).

@JosephWright VVVVV

@UlrikeFischer the babel manual explicitly says the shorthands differ from language to language. "- has been used for words with - for years, the - should be visible.

@PolineSandra it does not really belong there

@egreg Yes, that's come up I think already ...

@JosephWright Found and marked as duplicate

@egreg Cool

@daleif Javier is first person for babel these days, not sure who maintains the danish support

@DavidCarlisle danish.ldf mentions Johannes Braams, probably not up to date

@daleif I should assume that means Javier (or you:-)

If I use pdfTeX I can use \pdfximage and \pdflastximagepages to get the number of pages in a PDF file. How is this possible in XeTeX and LuaTeX?

@Skillmon LuaTeX: \saveimageresource, \lastsavedimageresourcepages, straight swaps for the pdfTeX names

@Skillmon XeTeX: \XeTeXpdfpagecount

@JosephWright thank you very much! I didn't know where to search for this :(

@JosephWright strangely I get 0 as output from \the\XeTeXpdfpagecount{example-image-duck.pdf}.

@Skillmon I've not checked the syntax/usage: quite possibly you need to load it first

\setbox0=\hbox{\XeTeXpdffile example-image-duck.pdf\relax}
\the\XeTeXpdfpagecount example-image-duck.pdf\relax
\bye

@Skillmon ^^^

@JosephWright the box is not necessary, the syntax was just wrong, I didn't expect the TeX syntax of file name inputting.

@Skillmon Ah, didn't check all the possibles :)

@Skillmon Only LuaTeX natively uses braced file names

@JosephWright in pdf(La)TeX \pdfximage{example-image-duck.pdf} works, too.

FEDORA CORE IS BACK

Should one write e-\TeX?

@PauloCereda better to use Word

@PauloCereda urbandictionary.com/define.php?term=Fedora-Core

@egreg ouch

@PauloCereda \mbox{$\varepsilon$-}\TeX

@egreg Thank you!

@PauloCereda use \hologo{eTeX} and leave the details to someone else

@egreg @DavidCarlisle I decided to not mess with it and leave the reference out. Thanks. :)

@egreg @DavidCarlisle @UlrikeFischer Sorry for bugging you here but has some memory size been changed from TL2017 to TL2018? (I am looking at this question and it seems that the core is running out of memory which it didn't with the previous TL distribution.)

@marmot will check.

@UlrikeFischer Danke!

Bitte!

ooh advanced German

@PauloCereda De nada?

@marmot ooh that's correct! :)

Marmots are way better with Portuguese than Englishmen. :)

@DavidCarlisle ^^

@PauloCereda I was hiking on Madeira... ;-)

@marmot ooh I went there, yay!

But that's not fair, everything in Madeira is a hiking expedition, considering the inclination. :)

@PauloCereda I was looking for other marmots there, no success....

@marmot oh no

@PauloCereda but they had good Madeira wine ;-)

@marmot I had none. :)

@PauloCereda You should try it.

@marmot @UlrikeFischer save stack, 99 times out of 100 that is a bug in the tex code

@DavidCarlisle I think that this time it is the 1 out of 100. The fillbetween library allows you to redraw intersection segments of paths, and if the paths have lots of points this costs lots of memory.

@DavidCarlisle I know. See tex.stackexchange.com/a/134599/2388 ;-). But the question is what changed between 17 and 18.

@marmot perhaps I'm struggling to even type at present as I'm building a few thousand xml files so response time is a bit slow so can't really debug the tex:-)

@marmot picture mode uses lass save stack than tikz

@DavidCarlisle Yes, and if you handwrite the document, you do not even need a computer.

@marmot Not that I can see.

@egreg And what does one have to do to increase it. (Sorry for asking all this.)

@marmot yes but tl2017 it uses 2036 save stack and 2018 the same libraries go over 8000 sounds like a bug to me

@marmot see the link Ulrike gave a couple of lines up has an answer showing how to increase

@DavidCarlisle OK, thanks. (I also just checked that the error also arises with TL 2017, so the OP made a wrong statement, sorry for not having checked before.)

@marmot it worked for me with tl2017

@marmot it works for me too with tl17.

@DavidCarlisle @UlrikeFischer I see. Interesting. What I did was just to use tikzlibrarypgfplots.fillbetween.code.tex. So it is not a change in that file but something else.

@UlrikeFischer @DavidCarlisle Sorry, wanted to say is " What I did was just to use tikzlibrarypgfplots.fillbetween.code.tex from TL2017. So it is not a change in that file but something else"

@marmot so I guessed, I just used /usr/local/texlive/2017/bin/x86_64-cygwin/pdflatex instead of pdflatex on the commandline

@marmot I just tried tl18 along with the previous pgfplots version and it worked fine. So some change from 1.15 to 1.16 is the problem.

{restoring \pgfmath@token@group@*@=undefined}
{restoring \pgfmath@token@group@^@=undefined}
{restoring \pgfmath@token@group@neg@=undefined}
{restoring \pgfmath@operator@neg=undefined}
{restoring \pgfmath@operator@@@=undefined}
{restoring \pgfmath@operator@)@=undefined}
{restoring \pgfmath@token@mathunit@)@=undefined}
{restoring \pgfmath@base@2)=undefined}
{restoring \pgfmath@token@unit@)@=undefined}
{restoring \pgfmath@token@group@^/=undefined}
{restoring \pgfmath@operator@/2=undefined}
{restoring \pgfmath@token@mathunit@/2=undefined}

@marmot looks like a bug to me

@UlrikeFischer I see. Thanks. But that seems to be very different from what @DavidCarlisle did, and on my machine setting \pgfplotsset{compat=1.15} does not make the error go away.

@marmot even with compat you are still using 1.16. I download the whole tds from 1.15 and attached it as a new tree.

@marmot At a certain point, the values of \dimen106, \dimen107 and \dimen108 start to be retained very often.

@DavidCarlisle I agree. Interesting! (When I encountered the same error message in some earlier attempts to answer a question, I thought that I was just using too complicated paths, and sure enough using simpler paths made the error go away.)

@egreg \pgfutil@tempdimb=\dimen106

@UlrikeFischer Also very interesting. (To me the way some of the paths get restored look a bit strange. If you think about, taking samples=100 is already a very high value, but this would only require one to remember 100 coordinates. But of course I have no concrete idea of how to improve on that.)

@DavidCarlisle I think that \global\somedimen=\somedimen is done quite often in PGF code. Like in \global\pgf@y=\pgf@y.

@egreg Yup

@egreg There's a reason l3draw works differently

@egreg Well that's a save stack waiting to be filled:-)

@JosephWright see starred comment on right

@marmot I guess the change is due to the new version of pgfplots.

It's pgfplots 2017/06/05 v1.15 in TL2017, and pgfplots 2018/03/28 v1.16 in TL2018

@egreg but @DavidCarlisle wrote "so I guessed, I just used /usr/local/texlive/2017/bin/x86_64-cygwin/pdflatex instead of pdflatex on the commandline" ???

@marmot The same I'm doing here, with x86_64-darwin of course.

@egreg So you are saying that if you run a pdflatex from an old distribution it also loads the old packages?

@marmot Yes

@marmot texlive binaries default TEXMFHOME from their location on disk, and everything else is specified relative to that, so old formats old packages etc

@marmot I found the commit that lead to the break sourceforge.net/p/pgfplots/code/ci/33425. But it changed quite a lot so imho better make a bug report ...

@UlrikeFischer @egreg @DavidCarlisle Thanks a lot! (I don't know how to make the bug report but I am going to recommend this to the OP if you do not mind.)

@marmot most people don't have old copies to hand I would guess

@marmot @DavidCarlisle The amusing thing is that pgfplots has fixed the many retained values of \pgfplotsretval in the fillbetween library…

@DavidCarlisle I thought @UlrikeFischer suggested to send a feature request to Christian Feuersänger.

@marmot oh sorry I thought you meant suggest run on an old texlive

@DavidCarlisle Naaah, marmots try to avoid running, especially old ones. ;-)

@marmot I have a suspicion that you are not a real marmot

@DavidCarlisle No, you are wrong. ;-)

@DavidCarlisle You just have to look at my profile picture.

@marmot it looks like a stuffed toy in a plant pot

@DavidCarlisle You seem just to be jealous because I have full hair. ;-)

@DavidCarlisle I am a real duck

@PauloCereda Strange, I should have thought you were an integer duck.

@PauloCereda Although ducks are generally floating, so maybe you are real after all.

@AlanMunn ooh Fortran joke

@AlanMunn LOLOLOL

I understand with TL2018 we no longer need [utf8]inputenc -- do we not need [T1]{fontenc} either, with pdflatex?

@musarithmia no, T1 is still needed.

@musarithmia No, you still need that. The default font encoding is still OT1

@UlrikeFischer @AlanMunn was getting T1 without asking for it. Maybe I was using a font package that called fontenc internally (e.g., Libertine).

@musarithmia Yes, that would likely do it.

@AlanMunn I always thought integers were real, really.

@PauloCereda really tasty

@DavidCarlisle oh no

@DavidCarlisle Even though one could find "Chickens, ducks and turkeys are great sources of protein that can help grow your hair long, thick and strong." in the internet I am not sure I would trust everything one can read there.

@marmot I don't believe anything that I read in this part of the internet

@DavidCarlisle Very good! And even better for the ducks!

Guys, what's the output of makeindex? An ilg file, right?

@PauloCereda no that's the log .ind is the generated index normally

@DavidCarlisle Ah thank you!

@PauloCereda but you can define other outputs.

@PauloCereda it's just a generated latex list with \item

@UlrikeFischer Got it, thanks!

@UlrikeFischer It's possible but a bit technical for a duck

@PauloCereda I had a document with four index. arara was quite useful there but I had to tweak it a bit:

% arara: makeindex: {style: stil.ist}
% arara: makeindexext: {style: stil.ist, log: slg, out: snd, input: sdx}
% arara: makeindexext: {style: stil2.ist, log: alg, out: and, input: adx}
% arara: makeindexext: {style: stil3.ist, log: vlg, out: vnd, input: vgx}

@UlrikeFischer I am working on this rule as we speak. Would you like me to enhance it?

@PauloCereda well currently I have no project with such indices ;-). And normally I would use imakeidx. But imho it should be possible to define such makeindex calls with arara.

@UlrikeFischer Got it.

@PauloCereda working on arara? I thought you'd be working on your....

@DavidCarlisle oh no

@UlrikeFischer: new rule does what you want. :)

@PauloCereda can I have a rule that makes it do whatever I want too?

@DavidCarlisle Yes, I thought of that. :) It will be the call rule.

@PauloCereda like this?:

% arara: call: {make: thesis}
\endinput

@DavidCarlisle ThesisNotFoundException oh no

@PauloCereda you have loads of time to write an arara auto thesis generation rule in time to defend it

@DavidCarlisle ooh automation

@PauloCereda it should be called "orange sauce" ;-)