How can I have a comma in Tikz label text (because comma is a deliminator I need to let Tikz know that I want to type it)?

Solved:

Sorry bothering, braces around the comma :)

@JosephWright I guess:

   \ifnum\outputmode>\z@\expandafter\edef\else\expandafter\def\fi \pdfcreationdate {\pdffeedback creationdate}

if we want to allow outputmode to change after the definition :(

@wipet ^^

@DavidCarlisle Yes, my thinking

Good maen

@JosephWright I did that but I'm wondering if {\outputmode=1\xdef\pdfcreationdate {\pdffeedback creationdate}} would be better so that its expansion behaviour doesn't depend on the way it was defined.

@DavidCarlisle Looks like it: pdfTeX defines the data even in DVI mode

@JosephWright OK I'll change it to xdef and push to ctan, they'll be pleased to see another update from me after yesterday:-)

I want to put two matrices, one above the other. I want the matrix columns to line up even between the two separate matrices. This means that each column should be wide enough to accommodate a single digit with a minus sign (even if no numbers in that column are negative). What is the simplest way to achieve this?

Should I use p{width} like here? tex.stackexchange.com/q/54959/6302

Or is there a something that simply says "at least as wide as -9"?

@Szabolcs dcolumn and a column spec of D..{2.0} would work?

@Szabolcs I would suggest using blkarray (@DavidCarlisle will chuckle)

@egreg your prediction re luatex85 package proved accurate, another update coming up

@DavidCarlisle :P

@egreg we broke csplain

@DavidCarlisle No

@DavidCarlisle Does csplain load luatex85.sty?

@JosephWright we broke csplain because we hadn't noticed it was misusing the package and getting lucky previously

@DavidCarlisle Still no

@egreg yes, before it switches to pdf mode....

@DavidCarlisle Bizarre!

@DavidCarlisle csplain is broken as it loads pdftexconfig rather than luatexconfig when used with LuaTeX and things that loading luatex85 is the 'correct' approach

@DavidCarlisle One might be generous and read it this way

@JosephWright gone to ctan

@JosephWright so out of 7 ctan uploads yesterday, one had spurious files, one had the wrong zip file attached and one broke a complete format. Not bad for a day's work:-)

@DavidCarlisle :-)

@DavidCarlisle I'm sure csplain with LuaTeX is very widely used

@DavidCarlisle Not clear to me that luaplain (csplain + LuaTeX) uses any pdfTeX primitives at all other than to load the shared config

@JosephWright I normally use it every day, but strangely not in the last week. ;-)

@JosephWright I suspect not

@egreg presumably because you've been too busy answering miktex questions

@DavidCarlisle I have troubles resisting the temptation of buying Windoze just to be able to switch to MiKTeX: the invaluable delights of fighting against administrator and user mode, of having to run updmap manually and so on!

@egreg I gather you'll be able to use miktex on osx before long (saw something about non windows builds somewhere I think?)

@egreg As @UlrikeFischer usually points out, for most users you need only a single user installation

@DavidCarlisle Several years ago I compiled mkm for Mac OS X and even tried to use it; it was before MacTeX was released.

@DavidCarlisle Yes, been on the list of 'to do' for some time

@JosephWright Yes, but users don't know it and install both modes, for “safety”.

This is a screen shot of TeX.SX window in Google Chrome on my mobile phone -- The badge dots are 'pacman' rather ;-)

Interesting, I thought csplain was used only by Czech and Slovak language authors.

@ChristianHupfer No pacman on iOS

What style of single quotes should be used inside \@latex@warning? I currently use `text`

@yo' Traditionally they are 'as for typesetting': `text'

@yo' Personally, I go with 'text'

@JosephWright ah ok, thanks

@JosephWright oh, so there's no strict policy on it :-)

@yo' Well latex.ltx is self-consistent

@JosephWright ok, I'll go with that then

@yo' the backticks are so annoying when you try to copy code fragments into here

@DavidCarlisle :)

And last thing: I suppose that tweaking \begin (even locally) is not the best idea, right? :) @JosephWright @DavidCarlisle

@yo' probably not, although I'm sure there are packages that do it

@DavidCarlisle my point is that inside my environment, & is active, but it can contain environments that need it to be catcode-4. I currently solve it by \ifx\@currenvir, which seems to work, and I'm not sure which is worse: the \ifx trick or tweaking \begin

@yo' do they really need catcode 4 or is an active & let to a catcode 4 one sufficient?

@DavidCarlisle well, I can't "let" it, I have to "def" it

\def&{%
	\ifx\@currenvir\eqna@currenvir
		\eqna@amp
	\else
		\eqna@tab@amp
	\fi
}

where \eqna@tab@amp is a catcode-4 & and \eqna@amp is some dirty stuff ^^ @DavidCarlisle

@yo' a couple of \expandafter in there would be safer

@DavidCarlisle like \expandafter\eqna@tab@amp\fi to make it expand after the conditional is finished? Oh right, because & ends a group and it could be a mess...

and the same with \eqna@amp since it sometimes expands to a catcode-4 & as well

so how's it correct then? this is right, right?

\gdef\eqna@newamp{%
	\ifx\@currenvir\eqna@currenvir
		\expandafter\eqna@amp
	\else
		\expandafter\eqna@tab@amp
	\fi
}

@yo' yes although as I suspected there are mis-behaved packages that hack begin, I just found this for example

\def\BA@blockarray[#1]#2{%
  \expandafter\let\expandafter\BA@finalposition
    \csname BA@position@#1\endcsname
  \let\begin\BA@begin
  \let\end\BA@end

but then I should probably have \def\eqna@tab@amp{&} and not \let, right?

@DavidCarlisle what is this blockarray? Because if they redefine \begin this blatantly (I mean, it's not patching, it's rewriting), this would break any \begin tweak done before

@yo' possibly, it depends,,,

@DavidCarlisle well, anyway, a full expansion (x or f-type) is needed, not o-type, so it proabbly doesn't matter

@yo' no idea, although since it redefines environments so they start in one group and end in another some re-writing might have been needed

@DavidCarlisle ok, I'll currently stick to the \ifx\@currenvir stuff and re-consider in case of troubles.

Where's my cake?

Writing this while in the shaky bus. :)

@PauloCereda ^^

@DavidCarlisle is that a cake?!

@yo' see the url: julietstallwoodcakesandbiscuits.co.uk/home/wp-content/uploads/…

@DavidCarlisle wow

is there someone expert on TikZ layered graph drawing system?

We are missing :-(

@DavidCarlisle ^^^ Flow chart gets me to where I think would be sensible ;-)

@JosephWright MIT?

@DavidCarlisle Certainly in that direction :-)

@PauloCereda Happy birthday!

@DavidCarlisle You broke bm ;-)

@egreg make yourself useful and fix it then:-)

@DavidCarlisle I guess it's the fact that luatex always returns \Umathcode and so under legacy fonts the code in bm fails. The fix is “don't use lualatex with legacy fonts”. ;-)

@egreg i left a grumpy comment

@DavidCarlisle You're becoming as grumpy as Brent.

@egreg well people shouldn't suggest my packages have bugs!

@egreg Indeed

@egreg it doesn't just fail with legacy fonts, it fails with opentype math fonts as well (consistency is a good thing) seeing as almost all of the package is concerned with unpacking the bits in a mathcode and resetting some of them to change the effective family, essentially none of it does anything sensible with a Umathcharnum

@DavidCarlisle the fix is easy: ``rm `which luatex```

@DavidCarlisle Yes, does seem somewhat tricky with LuaTeX (or presumably XeTeX + Unicode math font)

@DavidCarlisle Add to it the entertaining way of packing the code into a number

@yo' Tut tut, don't use which, you want command -v

On texlive 2016, fmtutil-sys fails to produce the luatex formats.

fmtutil [INFO]: --- remaking lualatex with luatex
fmtutil: running `luatex -ini   -jobname=lualatex -progname=lualatex lualatex.ini' ...
This is LuaTeX, Version 0.95.0 (TeX Live 2016/W32TeX)  (INITEX)
 restricted system commands enabled.
(c:/texlive/2016/texmf-dist/tex/latex/latexconfig/lualatex.ini
(c:/texlive/2016/texmf-dist/tex/generic/tex-ini-files/pdftexconfig.tex
! Undefined control sequence.
l.6 \pdfoutput
                       = 1
?
! Emergency stop.
l.6 \pdfoutput
                       = 1

@JosephWright locate luatex | xargs rm -rf

@yo' :-)

@KeksDose Nope, we fixed that

tlmgr update --all: no update available -- ?

@KeksDose Something is definitely wrong: the .ini file for lualatex no longer reads pdftexconfig.tex directly, you should see luatexconfig.tex first

fmtutil: running `luatex -ini   -jobname=lualatex -progname=lualatex lualatex.in
i' ...
This is LuaTeX, Version 0.95.0 (TeX Live 2016/W32TeX)  (INITEX)
 restricted system commands enabled.
(c:/texlive/2016/texmf-dist/tex/generic/tex-ini-files/lualatex.ini
(c:/texlive/2016/texmf-dist/tex/generic/tex-ini-files/luatexconfig.tex
(c:/texlive/2016/texmf-dist/tex/generic/tex-ini-files/pdftexconfig.tex))
(c:/texlive/2016/texmf-dist/tex/latex/base/latex.ltx
(c:/texlive/2016/texmf-dist/tex/latex/base/texsys.cfg)

@KeksDose Ah, you've got an old lualatex.ini in the wrong place

C:\Windows\system32>kpsewhich lualatex.ini
c:/texlive/2016/texmf-dist/tex/generic/tex-ini-files/lualatex.ini

How is a question closed? This: tex.stackexchange.com/questions/314160/… has 4 close votes, but the Review says "0 questions Close Votes". How do people know that a close petition is pending?

@Aditya Could you please have a look at this answer of mine? I ran into a problem there, when trying to define a new math operator. I have now used \definemathcommand[argmax][limop]{\mfunction{argmax}}, because the definition I wanted to use (after math-def.mkiv) \definemathcommand[argmax][limop]{\mfunctionlabeltext{argmax}} didn't work in the sense, that the text »argmax« did not show up in the output. Was I doing something wrong?

@Hans-PeterE.Kristiansen The review queue is funny. I've given up trying to understand how it works.

@JosephWright Sorry for not addressing you.

@KeksDose Odd

@egreg: OK. But logically that is where the info should be - it is not just me, that is missing something :o)

@Hans-PeterE.Kristiansen one problem is to tune what shall make it into the queue and what shan't. On a smallish site like this one you can argue that everything shall. But on larger sites, it can be different.

@Jos

@JosephWright I see, outdated version from 15-10-12, in comparison to the source. Which part of texlive should I reinstall?

@KeksDose I'd try latexconfig and tex-ini-files

@Hans-PeterE.Kristiansen The question is now on hold, but I can't see it in the history.

@Hans-PeterE.Kristiansen And I didn't see it in my queue when it still had only four closing votes

@egreg:I am satisfied by the answer "The review queue is funny" - now I know, that it is not just me that is too stupid to see what is going on. -also @yo' s explanation is making sense.

@Hans-PeterE.Kristiansen @yo' By the way, I keep seeing all the time the annoying brown square in the top bar telling me there are five items to review, just to find out that there's none (or maybe two).

@egreg AFAIK the number includes items in the queues which you've alreday reviewed

@yo' It typically happens a long time after I accessed the review queues: six items shown, no item to review.

@egreg -- there always seems to be at least one more than is actually in the queue. and, as @yo' already pointed out, if something is still awaiting review, but you've already looked at it (not just voted on it, but also said "skip"), it's still counted, but not visible.

@JosephWright So strange. I installed pretest and later simply swaped the pretest repo for the texlive-mirror, and upgraded --all. In the folder latexconfig remained color.cfg, dvilualatex.ini, graphics.cfg, xelatex.ini and lualatex.ini, all dated 2015-11-9, even after I removed the whole latexconfig (tlmgr remove --force latexconfig). However, deleted manually, reinstalled, worx. Thanks!

"How many yaks could a yak pack pack if a yak pack could pack yaks?" tonguetwister gone real:

@DavidCarlisle, @egreg: Thank you! <3

@PauloCereda Are you at home? Cake?

@egreg Sadly, I had to come to SP today. :( But there will be cake later on tonight. :)

@PauloCereda will it be bright yellow with a beak?

@DavidCarlisle I heard there will be chestnuts. :)

@PauloCereda Hope the shaky bus has no delay, or temptations at home would grow bigger

@PauloCereda oh chestnuts! Happy birthday, pal!

hello :)

aniversário do Paulo? parabéns

wondering, JabRef help page points to tex.SE but how off-topic a question about jabref gui appearence would be? i feel totally offtopic in asking that.

@G.Bay such questions are IMHO very borderline (if that makes any sense)

@PauloCereda Weil heute dein Geburtstag ist, da habe ich gedacht, ich singe dir ein schönes Lied, weil dir das Freude macht.

Happy Birthday

@PauloCereda Quack birthday ! :)

Now, being part of modern society using WhatsApp i realized: Punctuation marks and proper spelling seem to be out.

@Johannes_B Yes, this is apparently the case

@JosephWright :-(

@Johannes_B Indeed

This sentence no verb.

@StefanKottwitz code from texblog.net have a license or not? Unable to see anything about it

@RomainPicot Oh no, not license hell

@JosephWright Just thought the same :-)

@Johannes_B See my earlier post

@JosephWright Which one?

@Johannes_B Starred

@JosephWright ah, got it.

@JosephWright French and English word are sometimes really close :/

@JosephWright The daughter of a friend is 16 and her english is ... it could be improved. Is there any english/american TV show suitable for a 16yo? Personally, i love Green Wing, but this seems to be slighty inappropriate.

@Johannes_B Game of Thrones?

@yo' Isn't that one full of violence?

@Johannes_B may be, depends on your definition of violence.

@yo' Do you know the movie A history of violence?

Then, there are many American series. For British things, I recommend some of the crime series such as Midsomer Murders (there's plenty of these)

@Johannes_B for a british one I will say Doctor Who

@RomainPicot could be. I dislike that one, but it's probably good

@RomainPicot Feel free to use - code snippets there are so small that I did not even think about a license for code on my blog

@yo' I've lot of friend that like the show so it may be good (I'll watch it soon I think)

@StefanKottwitz thanks :)

@Johannes_B no, should I?

@yo' never heard of that.

@RomainPicot Never watched that.

@yo' Don't know. Was with that bloke from LotR and The Road. Interesting film, you should watch it some time.

@Johannes_B if she's into British humour and into crime series, she might like it. There's >10 series AFAIK

@yo' I asked her if she watches english stuff: No, don't want to. Wouldn't understand a thing. :-/

@Johannes_B English sound with German subtitles?

@yo' Subtitles are horrible for me. If there are subtitles, i read them. When they are in german, my brain goes: That one was incorrect. If the subtitles are in french or spanish, my brain reads them and thinks: Oh, interesting.

@Johannes_B :)

@yo' I learned some spanish chunks watching Breaking bad and Dexter.

@ChristianHupfer Had to fill in some la for the comment to be long enough.

@Johannes_B That approach of the O.P. is totally screwing up memoir

@ChristianHupfer didn't take a close look, but probably yeah

@ChristianHupfer By the way, there are no translations into other languages.

Learn proper inglizsh instead.

@Johannes_B It's a not very clever idea to define \newcommand\chapitre{\chapter} and assuming that \thechapitre and anything other command that slightly resembles the word \chapter would be defined in one go...

@ChristianHupfer I know

@Johannes_B Looks like Hungarian to me ;-) Time for the Hungarian-English phrasebook @PauloCereda ;-)

@ChristianHupfer A víziló táncoló spicc

@egreg My hovercraft is full of eels

@ChristianHupfer What French guy did that? :D

@yo' Don't ask me ... Instead of using \chapter directly he/she screwed up the memoir.cls with a bunch of definitions

@ChristianHupfer and published it as memoire.cls ?

@yo' Mais non. Maclasse.cls ;-)

@yo': Regardez ici ;-) tex.stackexchange.com/questions/314815/…

@ChristianHupfer holy merde

Please, don't you know who is latexstudio? github.com/latexstudio @JosephWright @DavidCarlisle @Johannes_B

@yo' What is that?

@Johannes_B someone who forked like a zillion of various latex repos, including latex3/latex3

@yo' No idea who that might be.

@Johannes_B no familiar followers either

I mean, my code is LPPL, so very close to WTFPL, but I don't want versions of the package lying all around the place...

@yo' Might be a guy from the CJK part of the world

@yo' Oh i know that feeling :-)

@Johannes_B yeah. But why in the world would they fork such stupidity as eqnalign?

@yo' People forked MDT-template. There is no limit on insanity

@Johannes_B :)

@yo' I like classicthesis because it is awesome. <- That guy needs to go to the hospital

:-)

@yo' I think this is the usual case: You need some kind of knowledge to see that something is crap.

@Johannes_B github.com/latexstudio/eqnalign/pull/1 (I changed the text a bit after posting)

@yo' The repos are not useful to anybody IMHO :-)

@yo' I can relate to that, and still ... there are users

@Johannes_B I mean, the packages are, the repos are not.

@egreg You know, I really don't like being put in a position of having to verify egreg answers (tex.stackexchange.com/questions/314923/…). They should be timeless and immutable like the Word of God, as we have come to expect over the years. What will be next? Emoticons?!

@StevenB.Segletes :)

@StevenB.Segletes Now I see the problem: the endnote is typeset twice. In the case of \footnote, the insertion does not migrate, in the case of enotez it is just stored and so it appears twice.

@egreg I also tried various things, but in the end settled for mechanizing my hack.

surely this has been asked before, but i can't find a good candidate. unless someone can identify one, i propose adding this one to the list of "often referenced questions". suggestions, please?

@StevenB.Segletes The fix is ugly as hell!

@barbarabeeton There is tex.stackexchange.com/questions/107967/… and tex.stackexchange.com/questions/35832/…, which are similar/related if not exactly identical.

@egreg ...but at least it is timeless and immutable. The earth has been restored to its axis.

@barbarabeeton tex.stackexchange.com/questions/301103/empty-table-of-contents/… seems like a more direct duplicate.

@StevenB.Segletes :)

@TorbjørnT. -- thanks. there are good points in both of those, but i really like the example used in the question i asked about. and i really like one of the comments by @ChristianHupfer that didn't make it into the answer, namely "TeX is a linear machine". so maybe i'll nag @ChristianHupfer to gussy up his answer to include that statement (and maybe pointers to the others), and put this one into the "often referenced" list anyhow.

@JosephWright Are you around by any chance, please?

@yo' No -)

@JosephWright ok. Good bye then :)

@yo' Yes: working on the dependency tree for our contributed tests

ok, I'll try to be clear and show some short code I struggle with

@Jos This is obviously wrong as \msg_error has to be _protected, but how would one then (conceptually) differentiate the two cases: (1) when I need the code to be expandable, and (2) when I care about the errors?

\NewDocumentCommand \ctufield { >{\TrimSpaces}m } {
	\ctuthesis_field_use:n { #1 }
}

\cs_new:Nn \ctuthesis_field_use:n {
	\cs_if_exist_use:cF { g_ctuthesis_field_#1_tl } { \msg_error:nnx { ctuthesis } { unknown-field } { #1 } }
}

@yo' If it's supposed to be expandable you should first of all use \DeclareExpandableDocumentCommand.

@HenriMenke yeah, that's another thing (I'm actually just working on a cleanup of all this stuff, I was ignorant to Expandable and _protected)

@yo' Also, it is not possible to issue an error expandably, because \message is primitive and hence unexpandable (only executable).

but still it doesn't solve the issue that ignoring errors yields typos and including errors prevents expandability.

@HenriMenke Not true, just nearly true :-)

@HenriMenke I'm aware of this

@JosephWright Then tell me the truth. I'm interested.

The problem is: Is there any way how to distinguish the two cases? I considered having an optional argument that would be the "default" -- this way the user could be informed that in order to be expandable, he has to provide a default value. But then, how about NewExpandable vs. New then?

@yo' Without seeing the use context it's tricky: really you shouldn't be trying to use a field in a context where you've not been able to check

@JosephWright the context is varying: sometimes the field is always set and contains a name of a language.

@HenriMenke You can arrange for an undefined csname to be set up in such a way that it provides error info

@yo' Just crash at the point where the field is expected to be set.

@HenriMenke Now I lost you :-(

@JosephWright goodness :)

@yo' You can just put \ERROR for all that the user cares: the run is going to have to be killed in any case

@yo' Suppose, you are using this to retrieve the author for \maketitle. You might make \author a preamble-only command to force setting it before \begin{document}, i.e. it has to be set at \begin{document}. If it is not set at this point you issue the error. If it is set, everything is fine and you can fully-expandably retrieve its value (unless someone tampers with the internals, which we don't expect).

@yo' See the kernel function for expandable errors

@HenriMenke but the problem is that the system is modular: it may happen that a 3rd person adds a new field that is necessary, and stuff.

@yo' Then design the interface such that a person adding a new field has to provide a default value.

I have to re-think this again and again until I see what's the best option. Because in some cases, I can't avoid expandability (like in \hypersetup) but in some other cases, I would like to have a double check and errors issues.

\documentclass{ctuthesis}
\ExplSyntaxOn
\DeclareExpandableDocumentCommand \ctufield { >{\TrimSpaces}m }
 {
  \ctuthesis_field_use:n { #1 }
 }

\cs_generate_variant:Nn \__msg_kernel_expandable_error:nnn { nnx }
\cs_new:Nn \ctuthesis_field_use:n
 {
  \cs_if_exist_use:cF { g_ctuthesis_field_#1_tl }
   { \__msg_kernel_expandable_error:nnx { ctuthesis } { unknown-field } { #1 } }
 }
\ExplSyntaxOff
\begin{document}
\end{document}

@HenriMenke Or that when you use it expandably, you have to provide a default value (or even just make a starred variant that uses empty value as default -- that actually doesn't sound that bad as an idea, the only problem is then whether it should be DeclareExpandableDocumentCommand or NewDocummentCommand

@yo' ^^^ That's what Joseph was referring to.

@HenriMenke I do this and @JosephWright kills me for using __msg :)

@yo' Forget about it, actually. I just read the docs: »However, the interface is similar, with the important caveat that the message text and arguments are not expanded, and messages should be very short.«

@yo' That is to say, the variant \__msg_kernel_expandable_error:nnx will not work as intended.

@HenriMenke x functions are always tricky. I think I'll go with the star vs non-star.

@yo' Also, I don't find default-initialising the fields so strange. Especially default initialising it with empty string or \c__xparse_no_value_tl.

@yo' If a person didn't set, e.g., the author why no put -NoValue- in his document?

@HenriMenke I would really prefer an error issued. I'll go with the star.

@yo' Just use plain \use:c { g_ctuthesis_field_#1_tl }. This way you will get Undefined control sequence which is an error.

@HenriMenke it's not that simple (IMHO, as I say, I now want to get into a working and valid state, and I may refactor this later on; I already considered that option, believe me or not :-) )

@yo' Or use a property list and use \prop_item:Nn \g_ctuthesis_fields_prop { #1 }. Then you will get an error akin to Property #1 not found which is more telling to a user than Undefined control sequence.

@HenriMenke prop_item is protected, isn't it? Ah no, it isn't, but it defaults to the empty string, which is what I can have for free anyway: \cs_if_exist_use:cF { bla_#1 } { }

@yo' You can use Lua for fully expandable errors.

\documentclass{article}
\usepackage{xparse}
\ExplSyntaxOn
\tl_new:N \g_test_field_exists_tl
\tl_gset:Nn \g_test_field_exists_tl { this ~ exists }

\cs_new:Nn \test_field_use:n
 {
  \cs_if_exist_use:cF { g_test_field_#1_tl }
   { \lua_now_x:n { tex.error("Field~\lua_escape:n { #1 }~is~not~there") } }
 }

\edef\temp{\test_field_use:n { exists }}
\show\temp

\edef\temp{\test_field_use:n { notexists }}
\show\temp
\ExplSyntaxOff
\begin{document}
\end{document}

@yo' ^^^

Recently, LuaTeX is not an option really :-(

@yo' Why not? Actually, recently LuaTeX is becoming more of an option than before.

@HenriMenke with all the bugs?! Yes, I wrote my thesis in LuaTeX from TL15. I wouldn't dare myself use the TL16 version.

@yo' I wouldn't call it bugs, rather backwards-incompatibilities.

@yo' I'm beginning to use ConTeXt and there LuaTeX 0.95.0 is very well integrated.

@HenriMenke well, currently, they appear as bugs. Whether it is labor pains or not, I don't know, but the fact that (AFAIK) the DVD is shipped with incompatible luatex and lualatex.ini (or whatever file it is) speaks for itself. (Not that I blame anybody, I just point out the obvious.)

@yo' by now the majority of issues are not bugs really just that luatex has diverged sufficiently that you can't just pretend it's pdftex, so packages need to update or do not work with luatex. there are bugs as well, and issues with under specified font handling, but still...

@DavidCarlisle well, that's your perspective. My perspective is that for all these reasons I'll stop recommending lualatex until it somehow settles down, and if you don't like saying "bugs", I'm fine with "features" :)

@DavidCarlisle Because of that, I want to apologize for accusing bm of having bug.

@DavidCarlisle Is it possible to explain shortly why bm misbehaves with recent LuaTeX?

@yo' Th eonly outstanding bug I'm aware of is with vsplit, but that didn't break any test in the latex2e test suite and I couldn't find a single real document affected, it broke one vsplit test in the latex3 test suite.

@DavidCarlisle yeah, but you know what I mean: if it's impossible to use such basic packages as bm, it means that unicode-math needs to be finished and used, despite I'll never want to input anything else than lower ASCII in math mode.

@HenriMenke oh yes easy: luatex unified \mathchar and \Umathchar so even if you use classic math setup like amsfonts if you use \meaning on a symbol instead of getting the classic tex mathcode hex value you get the unicode based mathcharnum value so teh bit battern is completely different. as 90% of bm package is about pulling apart that bit pattern to change the bits denoting the math fam to the bold family while keeping all the others, it is almost all wrong for luatex.

@yo' bm isn't the slightest bit basic it takes apart the \meaning of characters and twiddles with the bit patterns in the hex representation of the characters, the fact that bit patterns change if you move from an 8 bit system to a 21 bit one isn't exactly a shock.

@DavidCarlisle well, it's the basic in the sense that it's essential in pdflatex

@yo' I could make it work in luatex, I just need a free evening:-)

@DavidCarlisle I could offer you one, but it wouldn't work I'm afraid :-(

;-)

@yo' main reason for not doing it so far was luatex was a bit of a moving target in that area before christmas, and I'm not sure really what's the best approach, pdftex xetex and luatex are now all quite different in this area internally and it's not clear how much bm should or can hide that and offer the same interface on all three., or all 5 if you consider luatex+uniocde-math and xetex+uniocde-math which are different again

@DavidCarlisle bm for unicode-math seems a little redundant. After all there is \symbf and friends, which simply change the mathcodes of all the glyphs rather than obtaining and manipulating their bit representation.

@DavidCarlisle well, IMHO unicode-math should provide \bm macro on its own. My personal issue is that as I said, I prefer sticking to the macros for typesetting math, I don't need the unicode math ranges to work (actually, I don't like them because they are highly unsupported by everything). But I feel that lualatex without unicode-math is going to be unusable.

@DavidCarlisle Do you know the rationale behind merging \mathchar and \Umathchar?

@HenriMenke because internally luatex is using a unicode view of the world so the "classic" input is taken as a thin syntax wrapper making the same internal foms rather than having two separate math layout engines in the same engine. if you don't care about exact compatibility with tex (and luatex developers explicitly do not aim for that) then it is not an unreasonable position even though it is a pain for latex which has far more cross-engine and historical compatibility issues than context

@HenriMenke which is (now) luatex only and does not aim for legacy document compatibility in the way plain and latex do

@DavidCarlisle Judging from the commit history, LuaTeX development also seems to be driven mostly by the ConTeXt people.

@HenriMenke oh yes definitely.

@HenriMenke I wanted to note that, too :) It's easy to be compatible with your own product.

@yo' most of the unicode math fonts don't have a bold font so that makes bm difficult

@DavidCarlisle yeah. Well, there are the unicode slots, but they are unoccupied in the fonts.

@yo' no I mean that they don't have a bold font. so stix for example has the bold alphabets in the unicode ranges and unicode-math can reach them but it hasn't got bold summation or + or any other symbol as they don't have unicode slots they need a complete bold font

@DavidCarlisle oh rly

@yo' I believe that Unicode will always lag behind mathematicians always wanting new symbols and combinations of font features; I'd not be surprised if somebody distinguishes between normal summation and bold summation. Even TeX lags behind, for example not having a real support for “prescripts” and only a fixed set of types (some more would be needed).

I wonder how one does apply warranty for goods bought in the UK. (A steel watch bracelet stopped locking itself.)

@yo' it all depends (where you got it from and what the warranty states)

@DavidCarlisle 5-year warranty, I got it from amazon.co.uk, explicitly shipped to Prague (so I did not export it, they did)

Well, it's a small mechanical failure, but it could still easily cost tens of pounds repair.

The point is that the authorized service centre for the Czech Republic claims on their website that they only accept watches bought in the Czech Republic for warranty repairs.

@yo' amazon have a fairly good returns procedure usually, although not sure how it works from prague (I can just drop things off at a local drop-off point and they pick it up:-)

@DavidCarlisle it's too late for return, I have to claim the 5-year warranty (I hope claim is the right word here)

well, I sent a mail to their UK service centre and I'll see what they reply. Meanwhile, I may ask how much is the repair, maybe it'll be cheap enough that I sneeze at the warranty thingy for being too complicated.

@yo' yes but that presumably (could) involve sending it back for repair?

@DavidCarlisle yeah, probably at my cost (at least one way). I'd be happy if they agreed to pay for the repair here in Prague, it's round the corner :-)

@HenriMenke as i mentioned to @yo' \symbf gives you the unicode bold alphabet ranges but it doesn't give you (and in most unicode math font setups you can not have) a full bold symbol set like \bm{\rightarrow}