@DavidCarlisle See my comments now I've actually read the Lua code :-)

@DavidCarlisle We found some stray callbacks :-)

@JosephWright yes so I see

@JosephWright not yet, on team list?

@DavidCarlisle Yes

@JosephWright arrived now!

@JosephWright yes OK, I'm happy enough to go with it, I think the only place were you could do significant pruning would be to drop the user callbacks, but my lua is elementary at best so I'm not going to argue too strongly on lua design issues.

@DavidCarlisle Perhaps one to talk to Elie about

@JosephWright \luatexUcharmath :(

@DavidCarlisle But also as \Umathchar (you meant \luatexUmathchar I think!)

@JosephWright yes, but shouldn't we construct a table of luatex - umath before enabling the primitives (if we are prefixing)

Hi, does anybody understand this? Including comments

@DavidCarlisle Certainly can do

@DavidCarlisle lualatex-math and unicode-math both use the \luatexU... names

@DavidCarlisle Also mathastext

@JosephWright boo:(

@DavidCarlisle Both fixable quite easily but I don't know how long TL has had the raw \U... names

@JosephWright I think people shouldn't expect stability from beta software, I suspect we should just unprefix everything (perhaps or perhaps not) have a contrib package that puts the prefix back, and just force packages to update. luatex just doesn't have the same stability as tex82, I don't think we can apply the same rules to the latex layer over that, not yet at least.

@DavidCarlisle Well yes I agree but KB was worried that many end users don't really know any of that: they just want system fonts

@JosephWright and KB is right. Many people think "I'll replace pdflatex by lualatex and lmodern by fontspec and I'll be fine" :-)

@JosephWright sure but question is how best to achieve that.

@yo' Yes, but we can arrange that fontspec et al. do use the right names

@yo' yes but those people aren't writing half a million lines of package code setting up unicode fonts, so they are not the ones affected by low level interface changes.

@DavidCarlisle yeah I know, but they expect that everything works well in every version of TL and MT they install, which may be tough if you rename primitives :)

@yo' the number of people who will use a 2016 latex with a copy of unicode-math from 2014 is limited at best and if they really do that they can add \RequirePackage[2014/01/01]{latexrelease} and do that.

@yo' That's actually not the issue per se: Karl decided to rename the primitives so that they didn't clash with anything that the user had defined themselves

@JosephWright well, clashes with user macros seem to be yet another issue that I didn't mean at all, actually

@yo' One issue is we end up with people not being able to find the primitives when they do want them: they read the LuaTeX manual and try \localleftbox or whatever and it fails

@yo' We are very mindful of the needs of end users :-)

@yo' You'll note that for TL'15 we stuck with KB's approach pending further discussion

@yo' so the question is, is it really a good idea to confuse everyone who ever reads the luatex manual, just on the off chance that someone has a pdflatex document that has defined a local macro \localleftbox and gets a bad clash if they try that document on a lualatex that is using that name as a primitive.

@DavidCarlisle the only bad thing about confusing people who read manual is that you punish the handful of people who do the right thing.

@JosephWright yeah I noted that there's something like this going on

well, it's lunchtime for me, see you later

Good maen

@ChristianHupfer Quack!

@PauloCereda: From hence I call you "Paulo The Inevitable" :D

@ChristianHupfer Oh no!

@PauloCereda: Paulo the First? ;-)

@PauloCereda: Duck Vader? Better ;-)

@ChristianHupfer :)

@Arthur: Bonjour! :)

Hey @Paulo

@PauloCereda: I've blue-penciled 85 math graduation exams each having 3 larger problem sheets in about four days ... I hate math right now ;-)

@PauloCereda Quack!

@JasperLoy Quack! Hi Jasper!

@ChristianHupfer We all do. <3

@PauloCereda I am now using Fedora 22 Mate, with dnf instead of yum by default.

@JasperLoy Cool. :) How do you like it?

@PauloCereda It's fine, but I should uninstall Compiz because I don't need it.

@daleif: I think the memoir manual should be written in KOMA and the KOMA manual should be written in memoir. It would be great to make people mad. :)

@JasperLoy Oh. :)

@PauloCereda not even funny.

@daleif :(

Sorry.

@PauloCereda The KOMA manual might be a bit more readable though ;-). I know, I need to redesign and rewrite the memoir manual, but it is not high on my priorities at the moment.

@daleif I was just trying to make a light joke about two great classes. :) I like both classes (the arara uses memoir and some of my documents go with KOMA), but memoir still have the best documentation IMHO (I have to use the English version of the KOMA manual, which seems to not cover as much features as the German version does). I don't think they need rewriting, they are great. :)

@PauloCereda I think it is a bit clottered and general users may be able to understand it better if parts of it was rewritten or re-organized

@daleif Oh.

I wonder what is this supposed to mean :D

$10^4\,\rm{cm^{-3}}$

@yo' I guess it means (a) 10000 somethings per cubic centimetre and (b) the author doesn't know TeX that well.

@DavidCarlisle :-)

needed to say, it's a neat trick that I might use from now on, just without the extra braces (and don't call me dirty/naghty, please)

@yo' \rm ? it is of course entirely intentional that it has that effect in math mode, it isn't just some weird side effect.

@DavidCarlisle yeah I know, it's just that \rm is deprecated.

@yo' deprecated isn't really the right word, it's undefined by default:-)

@DavidCarlisle The \it syntax is inherited form LaTeX 2.09, and is regarded as supported 'for historical reasons only' in LaTeX2e. (quoting @Joseph from tex.stackexchange.com/a/517/11002 )

@yo' Yes, in say article.cls but not latex.ltx (perhaps I should make that clearer)

@JosephWright perhaps. Or maybe at least make it clear that it's valid (and to reasonable extent correct) to write $10^{7}\,{\rm s^{-1}}$ :)

@yo' It's valid 2.09, plain or ConTeXt

@yo' Try it in memoir.

@JosephWright but not in 2e? You two with @David always confuse me

@TorbjørnT. I don't quite use memoir :-)

@yo' as I say \rm is not defined in the latex format (hasn't been defined since 1993)

@yo' It's in for example article.cls so a fall-back from a 2.09 document will work

@JosephWright ok, but that does not answer the question whether it's wrong to use it.

@yo' it's wrong to use it in document classes that don't define it.

@yo' It's neither valid nor correct. It may work, but this depends on the class. Try memoir, for instance. If you have a paper where just a few units appear, use \mathrm or, better, define a shorthand.

@yo' whether it's OK to use in article depends. If article is the real target documentclass, then basically it's OK, if article is just being used as a generic MWE class, then the fragment may or may not work depending on whether the actual class defines \rm.

@DavidCarlisle ok, thanks.

@egreg well, my favourite (= the fastest to type) is $10^7\,$s$^{-1}$ :D

@yo' Which is, of course, wrong.

@yo' I'm just so going to add \mathsurround=1in to every document I get from you.

@ChristianHupfer not palindrome: fail.

@yo' Try it in a theorem statement.

@DavidCarlisle For me, there is no need for a palindrome... I just remember the song :D

@DavidCarlisle it's a pseudo-palindrome, and it's a morphic image of a non-trivial palindrome, good enough for me :)

@egreg I know. I'm not paid for being right, I'm paid for producing the right output

@yo' Too kind ;-)

@egreg those people who use units rarely have any theorems. I think I haven't seen any single such paper

@yo' \NewDocumentCommand{\un}{sm}{\IfBooleanF{#1}{\,}\mathrm{#1}} and $10^6\un{s}^{-1}$ is faster, isn't it?

@yo' And \SI{10^6}{s^{-1}} is even faster.

@egreg remove the spurious braces and it's the same :)

@egreg no, it is not, if you count keystrokes

@yo' Counting keystrokes is exactly the worst thing to do. You want the typescript to be clear, rather than saving a couple of bytes.

@egreg it's a conversion from W@#$, any attempt for clarity fails by default.

@egreg which situnix version supports that? Invalid token in number ;-)

@daleif Yes it should be \SI[parse-number=false], but it was just to show the idea.

@egreg I would perhaps have written it as \SI{e6}{\per s}, not sure of the readability (though)

@daleif I have a strong feeling that $10^6\,$s$^{-1}$ starts to be more readable than that :-)

@yo' urgh

@daleif it's a matter of habit. Also, the input I have is (modulo superscripts and substripts that get ignored by copy-paste): 106s-1

@yo' If your numbers with units are of that form, you just say \sisetup{parse-numbers=false} once and for all.

on a slightly simpler note: I wonder what is, according to the authors, \begin{figure*}[tbh!] supposed to do in a 2-column document :-)

@yo' One never knows, maybe @DavidCarlisle has finished xor in the meantime.

@ChristianHupfer No need for the extra version. :D

@Alenanno Too late :-P Please note the number of items now :D

@ChristianHupfer Ehehe

@egreg :D

@egreg perhaps it is finished already, we just don't tell you about it.

@DavidCarlisle @egreg has an earthshaking announcement in this year's TUG, maybe he has finished xor for you. :)

@ChristianHupfer I am always clicking on the symbol (which is dark grey on light grey for me), because that stupid +int annoys me so much.

@PauloCereda it'll be rubbish

@PauloCereda speaking of whom, did you notice my recent git commit log:-)

@DavidCarlisle ooh let me check

@PauloCereda faker

@DavidCarlisle lalalala can't hear you

@DavidCarlisle LOL best commit message ever

@Johannes_B Perhaps you should attend a therapy meeting "How do I stand getting reputation" :-P

@ChristianHupfer I think i didn't answer a single post today, neither here, nor golatex, nor LC, nor texwelt, now mrunix nor ... left two comments, though.

@DavidCarlisle Where is that git repository? I’m curious too :-)

@ArthurReutenauer I'm sure it's a secret

@ArthurReutenauer No one knows :)

@DavidCarlisle Thanks, found it :-)

catsbeaversandducks.tumblr.com/post/122176035941/…

@PauloCereda I burned TL2015 ISO to my USB stick and twice it got corrupted. What can I do?

Maybe I should change the stick. It installed fine on one machine and then by the time it got to another machine it got corrupted.

Then I redownloaded and reburnt and the same thing happened.

@JasperLoy Did you run the checksum?

@PauloCereda It seems it gets corrupted after burning, not before or during. Like a long time after.

@JasperLoy You could try mounting the ISO instead of burning it.

@PauloCereda I see. Can insufficient RAM cause untarring to fail during installation? I suspect that might be another factor!

@JasperLoy Good question. I really don't know. :(

@PauloCereda OK. I thought you knew everything. =)

@JasperLoy: Where do you want to burn the ISO file to? As a raw image to the USB stick file system?

@ChristianHupfer Yes, I bunrned it the same way I burn a Linux ISO. It installed fine twice on X but not fine twice on Y, and when I checked the image got corrupted (supposedly) somewhere in between.

@JasperLoy: How do you do this? Pure dd if=texlive.iso of=/dev/your_device_id with some appropiate block size?

@ChristianHupfer Actually, there are two methods I use, and both work (for X). cp a.iso /dev/sdb, or use the "restore disk image" function in the GNOME disk utility application if you use the GNOME desktop in Linux.

@JasperLoy GNOME is unknown to me ;-) And I've never seen that cp works with devices. Either I missed this or this is a relatively new feature. I've done such stuff with images (unless writing to a CD/DVD) with the good old dd tool. I rip images with dd too

I am very surprised so many people use dd when the above two methods work just as well and is documented for installing Linux!

I learnt about cp from the Debian website and GNOME disk utility from the Fedora website, so they must be correct, and they do work!

Now i see chat flags and can't even check for context? Kopfschüttel

Oh new image uploader, now supports drag and drop in firefox...

@DavidCarlisle Finally, more of your drawings, nicer, quicker uploaded then? ;-)

@ChristianHupfer ^^^

@ChristianHupfer chat still seems to have the old uploader

@DavidCarlisle I am flagging this an insult ;-)

@DavidCarlisle I referred to the normal posts, not to chat...

@ChristianHupfer @Johannes_B was asking for more chat flags.

@DavidCarlisle I don't like it. It means yet another click for every image upload and uploading images already required an excessive number of clicks, IMNSHO.

:22318930: Did not test it so far. I use Linux without any file manager being open, so drag and drop should be done within a pure Firefox file://directory listing

Can somebody please test the following example and tell me what they see?

% arara: lualatex
\documentclass{article}
\usepackage{libertine}
\usepackage{listings}
\begin{document}
\begin{lstlisting}
abc
def
xyz
\end{lstlisting}
\begin{lstlisting}
abc

лилиа
ελενικκι
\end{lstlisting}
\end{document}

@Johannes_B: I suspect Cyrillic or Greek letters, but since it is inside a listing...

@ChristianHupfer You should not use a colon there ;-)

@Johannes_B Ah, yes... stupid error, thanks

@Johannes_B It's known that listings doesn't know what to do with characters above ASCII 127

@Johannes_B I'm a bit surprised about the missing newline, but that's all.

@egreg I didn't, so far none of my listings had any non-ascii chars.

@egreg Yes, me too.

@egreg I didn't know too, but I did not program with other letters than coming from the ASCII code. I wonder how Russians, Chinese, Greek do this then...

@ChristianHupfer minted?

@egreg Not used it so far. But good to know

@egreg Can you please add ä Ü under the greek line and test again?

@Johannes_B If you remove the line breaks in the second listings the result is even funnier

@Johannes_B I just ignore the flags. It's mostly children playing.

@egreg Got it. Funny indeed. wait, funny?

@egreg If i am not completely mistaken, Ä and ß are also >127

% arara: lualatex
\documentclass{article}
\usepackage{libertine}
\usepackage{listings}
\begin{document}
\begin{lstlisting}
abc
ελενικκι
лилиа
Ä ß
\end{lstlisting}
\end{document}

Have a nice time... it's time to go to bed. Have slept only 4 hours last night.

@Johannes_B Possibly the limit is 255

@Johannes_B If I try Ā (U+0100), I find the same behavior as for Cyrillic and Greek. Instead ÿ (U+00FF) is regular.

@egreg Thanks for analyzing. Just skimming through the manual to say any restriction mentioned.

@Johannes_B but that document doesn't have an encoding specified so it's effectively latin1 with everything up to 255 being 1 byte (and the Greek mis-encoded) if you specify utf8 then only characters up to 127 are single byte

@DavidCarlisle How would i specify utf8 for listings?

@Johannes_B well either \usepackage[utf8]{inputenc} and listings understands, or you can't.

@DavidCarlisle I can't seems to be a bit unsatisfying.

@Johannes_B there is certainly no chance of pdflatex interpreting it as utf8 on its own.

@DavidCarlisle listings cannot process utf8 with pdflatex, thats why Heiko introduced listingsutf8 which does somewhat of an iconv along the way. I thought all that chaos would be done with LuaTeX and XeTeX.

@Johannes_B you shouldn't expect a system built on a 1byte=1 character model to be able to typeset the full unicode range. The fact that [utf8]{inputenc}

@Johannes_B but that isn't a luatex document is it? (no fontspec etc)

@Johannes_B I was assuming pdftex

@DavidCarlisle If run with LuaLaTeX, then it is. libertine tests for an engine and loads it.

@Johannes_B oh that funny arara comment thing should have given me a hint, I never use @PauloCereda's code:-)

@Johannes_B ah then listings could work over the full range (whether it does or not is a different issue) I suppose I could try your document....

@DavidCarlisle Proud user of latexmk ;-) But i know there is an arara rule for latexmk

@DavidCarlisle :-)

@Johannes_B that's weird the letters come through but line endings lost?

@DavidCarlisle You should have seen my face when i first watched the output. ;-)

@Johannes_B ah I think I see what's happening

try :

\documentclass{article}
\usepackage{libertine}
\usepackage{listings}
\begin{document}
\begin{lstlisting}
[a][b][c]
[ε][λ][ε][ν][ι][κ][κ][ι]
[л][и][л][и][а]
[Ä][ ][ß]
\end{lstlisting}
\end{document}

@Johannes_B listings really doesn't want you to use unicode, it makes all letters active initially defined to insert themselves, but can be defined to do other things, for syntax highlighting, but all letters is:

\lst@CCPut \lst@ProcessLetter
    {"40}{"41}{"42}{"43}{"44}{"45}{"46}{"47}
    {"48}{"49}{"4A}{"4B}{"4C}{"4D}{"4E}{"4F}
    {"50}{"51}{"52}{"53}{"54}{"55}{"56}{"57}
    {"58}{"59}{"5A}
         {"61}{"62}{"63}{"64}{"65}{"66}{"67}
    {"68}{"69}{"6A}{"6B}{"6C}{"6D}{"6E}{"6F}
    {"70}{"71}{"72}{"73}{"74}{"75}{"76}{"77}
    {"78}{"79}{"7A}
    \z@

@Johannes_B which is a rather short list of letters by unicode standards.

@DavidCarlisle And would be slowed down a tiny bit if the full Unicode list is supplied.

@egreg catcode table for luatex?

@DavidCarlisle Personally i wouldn't need that, but would there be anything to do about it? As you mentioned catcode tables?

@DavidCarlisle Not sure it would do.

@Johannes_B (@egreg) I don't know listings well enough, it would be easy to extend its initial setup so all characters were active (as egreg says that may have a time cost but possibly not too bad) but then I suspect assumptions about single byte encodings spread through the rest of the code for setting up the language highlighting, so it may be a bigger job than I would wish just now...

@DavidCarlisle I never dived into that area before. The manual describes some escaping.

@Johannes_B I suspect the manual isn't going to help

@DavidCarlisle Should we ping the maintainer?

@Johannes_B something like "dear maintainer there is a new(ish) and not totally stable tex variant would you rewrite your package to match?"

@DavidCarlisle Not quite, more of Look, surprised me. Do you think there can anything be done about it in the future? Any ideas?

@Johannes_B I suspect for Unicode a different approach might be better. Get @egreg to write a package using l3regex

@Johannes_B same thing:-)

@DavidCarlisle Now that you mention it, would that be something L3 would consider as kernel stuff?

@Johannes_B current l3 code is really only programming layer but if it doesn't provide enough to write a unicode aware syntax highlighted code package then the code layer isn't powerful enough so it's a valid use case for checking teh design but you wouldn't expect expl3 to support it out of the box

@DavidCarlisle That might be something to consider for 2028 i guess ;-)

@DavidCarlisle That would be an idea. Maybe for 2020.

@DavidCarlisle :)

@egreg: I am sure you saw this: github.com/davidcarlisle/dpctex/commit/… :)

@PauloCereda never trust a duck to keep a secret

@DavidCarlisle Quack! :) We are good at secrets, we just have a short attention span. :)

@PauloCereda @DavidCarlisle Cwac! I'm wondering when TL might get the update ;).

@cfr ooh :)

@PauloCereda Shame on you!

@PauloCereda I can drop etex from my forest trees in standalone class once the update comes through. (I could install it locally, I know. But then I'll forget it and miss later bug fixes ;).)

@cfr yes I expected it today. I'll leave it another day then ask if it doesn't arrive, moving from ctan to TL isn't automatic it requires manual volunteer work to check things in to the tl sources, so it's not good to look too impatient, but occasionally things do just get missed so at some point I'll ask.

@cfr :)

@PauloCereda The update is all @cfr's fault you know.

@DavidCarlisle :)

@DavidCarlisle Yes. I know how it works. I did a little of it for a bit ;).

@cfr actually I said it didn't need to go to ctan-announce on the upload form, I didn't think that would prevent Karl et al picking it up, but I wonder...

@DavidCarlisle Hmm.... I don't know. It seems unlikely that would matter as it would be something of a flaw in the system. I doubt the scripts are reading CTAN announce, anyway ;).

@PauloCereda I sent him a keyboard which inserts spurious spaces when you aren't looking.

@cfr oooh I like the plan. :)

@cfr that's what I thought, but when it dodn't appear I wondered...

@PauloCereda it would be much easier maintaining packages if people didn't use them

@DavidCarlisle You don't like the method of saying everything people complain about is a feature, then? Seems to work for the folks responsible for major Linux desktops. Or else, if a feature disappears they say that the people using it are not in their target audience.

@DavidCarlisle You heard me saying the very same thing in my talk in last year's UK-TUG meeting. :)

@PauloCereda So plagiarism as well?

@cfr oh not at all, I mentioned that when people started using arara, I started to have questions about it, and I had to answer them. :)

@PauloCereda obvious questions like "why isn't it written in emacs lisp" for example

@DavidCarlisle :)

@David: you know I made arara Cygwin-friendly just because of you, right? :)

@DavidCarlisle I guess if it is not announced, it might not leap out at people as high priority. Normally, there are a bunch of packages and you just pick the ones you are going to work on. At least, that's how it worked when I did it. (There are some which require more expertise so I ignored those, but there were almost always a bunch I could do and only so much time....)

@David: github.com/cereda/arara/issues/62

@PauloCereda only in a mythical future verfsion though?

@DavidCarlisle The so awaited 4.0 version. It seems you and I will ship xor and 4.0 at the same time. :)

@PauloCereda I wasn't serious ;). Ducks have no sense of humour....

@cfr oooh

@cfr: catsbeaversandducks.tumblr.com/post/122176035941/…

@PauloCereda ;)