@moewe. Following up on your comment at this question. When are the delimiter contexts (citecommand, bib, biblist) set in the standard biblatex styles? In your comment you imply that anything that calls \usedriver will be in bib context, is that right? I would prefer to have the cite:… drivers still use the citecommand context. \printbiblist also ends up calling the main drivers. Is the delimiter context there also overridden?

@DavidPurton Usually all \...cite commands set the delimcontext to their name (so \parencite sets the context to parencite etc.), \printbibliography sets it to bib, \printbiblist to biblist. \usedriver also sets the context to bib. The context assignments are basically \edefs, so they are local and the last one wins. You can print \blx@delimcontext to see which context is active if you are unsure.

@DavidPurton Depending on what exactly you want for your cite:drivers you could either set the delimcontext explicitly with \delimcontext{citedriver} at the beginning of the driver or you could use the explicit optional argument as you did.

Maybe we should consider changing the order of execution in

\protected\def\blx@imc@usedriver#1#2{%
  \begingroup
  \csuse{blx@hook@usedriver}%
  \csuse{blx@hook@bbxinit}#1%
  \delimcontext{bib}%
  \blx@beglangbib
  \blx@driver{#2}%
  \blx@endlangbib
  \endgroup}

to

\protected\def\blx@imc@usedriver#1#2{%
  \begingroup
  \delimcontext{bib}%
  \csuse{blx@hook@usedriver}%
  \csuse{blx@hook@bbxinit}#1%
  \blx@beglangbib
  \blx@driver{#2}%
  \blx@endlangbib
  \endgroup}

so that one can override the context with hook code or the precode of the \usedriver invocation.

@DavidPurton github.com/plk/biblatex/commit/… implements the change suggested above. You can now override the delimcontext with the precode \usedriver{\delimcontext{citecommand}}{cite:\thefield{entrytype}}

@DavidCarlisle Best not to dwell on that number just at the moment… Perhaps we will get better numbers later today. If it doesn't rain. Although I heard it always rains where you are.

@DavidPurton our cricket expert, @PauloCereda, assures me that England will take the ashes this year.

@DavidCarlisle We shall see :).

Morning al!

Been thinking about

I'm asking the Powers if it's doable

I still think it's the long-standing issue: most users don't need to be in a user group ...

Speaking of which: @DavidCarlisle, I guess I need to book The Dungeon

@JosephWright I must admit I'm not sure what the purpose is, it will just highlight that most users are not members, and as you say, it's not clear why most tex users would want to be a member.

@DavidCarlisle I think there's a reluctance to accept this from a range of people who've been involved for a long time; they see what TUG or UK-TUG or ... was 20+ years ago, or perhaps think that DANTE is a model we could all achieve (and DANTE is shrinking too)

@DavidCarlisle I think this year's TUG meeting was useful due to the attendance of some 'outsiders', but even for developers, focussed meetings are likely better today than a general 'get together'

@DavidCarlisle Let's be honest, UK-TUG is 'free pizza plus meet a few friends', plus 'have the dratted AGM'

@JosephWright I only go because you hold it in Oxford

@DavidCarlisle I hold it in Oxford so you will come ;)

@DavidCarlisle Free pizza?

@JosephWright What about DANTE's model is different from the other user groups?

@JosephWright well it costs the bus fare in and car park:-)

@schtandard having enough users to be worthwhile and enough language-specific issues that a local user group makes sense

@DavidCarlisle :)

@schtandard They hit a critical mass where they have local meet-ups for a talk/drink around Germany, so they have like 2k members and the UK group has <100!

@HenriMenke I think the context of your \begincsname question was ... pretty obvious :)

@DavidCarlisle Well then the solution is clear: We just have to force everyone to write in German, then every User Group can benefit from those language specific issues.

@JosephWright I didn't realize other User Groups were that small. TUG is better off, I assume?

@schtandard Er, smaller than DANTE

@schtandard There's a reason for the concern: we could well end up with just a small number of people paying for the CTAN hosting, etc.

@schtandard guter Plan

@JosephWright if tug wants money they should ask for 1 cent per user from overleaf etc, makes more sense than trying to get members to sign up

@DavidCarlisle I think Overleaf and others probably would be willing to prop up CTAN hosting and the like; it's actually cheap and is important to their business

@JosephWright I would have thought DANTE contributes to those payments.

@schtandard DANTE pay for CTAN

@DavidCarlisle tug.org/instmem.html

@DavidCarlisle Well, the last time Germans tried to impose themselves on the rest of the world did not go so well, so maybe someone else has to take the lead on that.

@JosephWright yes I know but that's more geared towards some large university department chipping in a bit as a goodwill gesture not a company with millions? of users and for which tex is the core buisness

@JosephWright Good.

@schtandard Boris J?

@DavidCarlisle Yes

@DavidCarlisle Like I said, actually paying for the hosting would probably be appropriate. That's what happens with e.g. CPAN: a company supply the core node

@DavidCarlisle Why not. He seems to be popular. Be sure to pitch it to him next time you meet. That would maybe take some focus off the whole Brexit business as well.

@JosephWright ooh

@DavidCarlisle Shareware LaTeX

@PauloCereda pay to get a premium version of longtable with fewer features...

@DavidCarlisle and thus less bugs? :)

@PauloCereda there are no bugs

@DavidCarlisle ooh

@JosephWright ooh new commit

@JosephWright Well, Hans doesn't browse TeX.SX, although once in a while he does because at some point he emailed me about an answer I wrote here.

@schtandard The current imposition of Germany on the world in form of the EU is going fairly good, I think.

@JosephWright imho both types are useful. The main point is that sometimes you should meet persons in real, one can't do everything only over distance.

@HenriMenke oh, I just found the reference. So you tried that yesterday too ;-)

@UlrikeFischer link?

@DavidCarlisle Henri asked on the context list, obviously in connection of the question yesterday about the impact of local/global csname-def on the save stack.

@DavidCarlisle mail-archive.com/ntg-context@ntg.nl/msg91847.html

@UlrikeFischer I thought that would be it but didn't see anything on dev-context archive, ah ntg-context...

more people should use xmltex

@DavidCarlisle ooh

@DavidCarlisle so that you get more feature requests?

@MarcelKrüger I just merged next-dev into dev as discussed and will push soon, just need to change the remote names ...

@UlrikeFischer Nice

@MarcelKrüger can I delete the next-dev (which I used only for the tests)? oh probably not, there is a pull request. I should really try to remember this ...

@UlrikeFischer oh no

This sounds like linked lists :)

@UlrikeFischer I changed the PRs, deleting should be fine now.

@MarcelKrüger oh no

@MarcelKrüger well you have write access ...

@UlrikeFischer Oh no, I no longer have an excuse to delegate all the work to you.

@MarcelKrüger oh no

@JosephWright & other LaTeX people: When I need additional data files in unicode-data, should I just push them to dev or should we use Pull Requests first to allow some discussion?

@MarcelKrüger I think we are trying to move towards a model where we use PR more

@DavidCarlisle how do one make a PR without external fork? With a branch?

@UlrikeFischer yes it's just a wrapper around a diff between branches

@DavidCarlisle I should try it ;-)

@DavidCarlisle the latex testsuite doesn't like my luatex - lots of differences kern0.32639 against \kern 0.32639 ;-)

@MarcelKrüger This is pretty uncontroverial: can just go in

@UlrikeFischer oh that space comes and goes with luatex releases..., I should check here but I'm travelling later so not a lot of time today, need to practice my German...

@DavidCarlisle ;-) not really a problem, I will simply ignore it. Where are you going?

@UlrikeFischer Zurich, Frieburg, Zurich, Innsbruck, more or less. Good job I'm fluent in the language.

@DavidCarlisle but you need to practice the ü of Zürich (and I hope you know also switzerdütsch).

@UlrikeFischer minor details:-)

@JosephWright how do we activate travis on luaotfload? Marcel pushed a new branch but it obviously didn't run the tests.

@UlrikeFischer you should be able to log on to your travis account and activate the repository from there.

@UlrikeFischer I just flicked the switch at travis-ci.org/organizations/latex3/repositories

@DavidCarlisle I see only an error page there, and on my travis account the repo is there but the branches are no up-to-date.

@UlrikeFischer odd, still I enabled travis for latex/luaotfload so if the travis yml file is there it should run on the next commit

@DavidCarlisle I will simply try to make a small change.

Had chocolate!

@JosephWright did you had any reaction about the filehook business? all my tests with just failed because of this ;-(.

@UlrikeFischer oh yes I fixed that locally I think last time I ran some tests

@DavidCarlisle i have for now added it to luaotfload locally too. But it is a bit of nuisance.

can someone tell me why my headers are not aligned?

@UlrikeFischer @JosephWright ^^^

@sollniss your cell entries are in p or l so flush left and your headings are centred

@DavidCarlisle where do you see this?

@DavidCarlisle even when i change it to \begin{tabular*}{\textwidth}{llllll} the headers are still centered somehow

@UlrikeFischer I forced a build based on the last checkin and it shows up travis-ci.org/latex3/luaotfload/builds/573134563

@sollniss yes your thead macro will be doing that

@sollniss if you use tabular* always use extracolsep (but personally I'd use tabular, forcing that columns apart just makes the table harder to read)

@DavidCarlisle hm, it shows only one branch.

@DavidCarlisle i found the layout a while ago and i'm not sure i remember what all parameters do. so you are suggesting either keeping the \extracolsep or changing tabular* to tabular?

@DavidCarlisle yes just saw it.

@sollniss yes always, if you use tabular* then you are specifying a target width for the table, but \extracolsep is the only mechanism to achieve that width so if you don't set it it will stay as 0pt so the table has unachievable constraints and will always be underful.

@DavidCarlisle ooh

@sollniss none of this is related to the table headings being centred though

alright thank you. i'll figure out how to align the headings differently and play around with it a bit

@sollniss you may prefer to use the S column defined by siunitx rather than use \SI in each cell.

@DavidCarlisle i'll look into it, thank you

another unrelated question, is it possible to define something as "uninterruptable" (if that makes any sense)? i have an enumerate which gets interrupted by a figure being placed in the middle by latex

A historical question. Did ConTeXT start using PDFTeX well before LaTeX? Because my memory of looking at ConTeXt back in the day was of this shiny new thing called PDF, which I wasn't familiar with. That was when LaTeX was still using DVI and dvips.

@sollniss you can try putting \suppressfloats[t] at the end of the enumerate so that (in theory, it doesn't always work) there can be no top floats on that page

@FaheemMitha no

@FaheemMitha latex still makes dvi by default. but there was a pdflatex from the earliest days of pdftex (I think at one point the pdftex email list was just was Thanh, Karl, Sebastian and me.)

@DavidCarlisle ooh a trilogy

@PauloCereda with four people ...

@UlrikeFischer oopsie

@UlrikeFischer we ducks are good with advanced math. :)

@DavidCarlisle What ConTeXt always based on PDFTeX, though?

@FaheemMitha not sure, I suspect not. I never used it much.

@DavidCarlisle Yes, I guess it's still a default, but doesn't everyone use the PDF option these days?

@FaheemMitha ask @barbarabeeton :-)

Well, nearly everyone. Does anyone do polls?

@FaheemMitha I was just looking on wikipedia but the dates are fuzzy but I think context mk1 is older than pdftex

@FaheemMitha also anyone using pstricks psfrag etc is probably using dvi, the japanese ptex variants produce dvi, as of course does xetex

@DavidCarlisle XeTeX can't produce PDF? I seem to remember it doing so.

Though I've only used it a little, and not very recently. Last March, I think.

For writing Devanagari. Which I believe is still currently better done in XeTeX.

@MarcelKrüger I fighted a bit with the "dev"-stuff in your new branches. I would like to be able to use the ctan luaotfload with lualatex and the dev version with lualatex-dev so that I have fast access to both. I can rename the branch locally to ignorable-dev to force this, but it gets a bit messy with the texmf. Do you have any suggestion how to sort this?

@FaheemMitha it makes xdvi (slightly extended dvi) then runs xdvipdfmx to convert that to pdf, so it looks as if it makes pdf, but it doesn't (so to packages like graphics/color/tikz etc that need to know what the back end is doing it is far more like tex thanit is like pdftex)

@DavidCarlisle I see. I didn't know that. So it's not based on PDFTeX then.

@FaheemMitha yes although harftex or luahbtex should be able to finally set Devanagari with luatex.

@FaheemMitha no it's a Unicode extension of etex. not of pdftex

@UlrikeFischer I just realised the same problem, after trying to figure out why l3build install didn't work correctly. You can just rename the branches on GitHub.

@DavidCarlisle I see.

@MarcelKrüger so you are ok with naming everything dev? Good. We should then delete the texmf/tex/luatex/luaotfload file so that only tex/latex-dev is there. I will update the branch (I also added temporatly a corrected filehook and the PropList.txt-file)

@DavidCarlisle -- Your mention of xmltex reminds me ... at the meeting, one slide showed an example of content mathml. Is there a document somewhere showing what is defined, or howq to define new concepts? (The example seemed plausible; I think it's the first one I've ever seen. Chris once told me he didn't think it was possible. I failed to ask him about this example, or about the current status.)

@barbarabeeton mathml spec lists all the content elements w3.org/TR/MathML3/chapter4.html you define new concepts vis <cymbol>bessel</csymbol> etc you can't define new elements such as <bessel/> but that's just syntax. We are working on mathml4 if you are interested... w3.org/community/mathml4

@barbarabeeton who's talk showed content mathml?

@FaheemMitha -- As of when I retired from the AMS in February, they were still generating dvi for proofing the print version of documents -- it's much easier to automate graphics checking by that route. After proofreading and corrections, the pdf route is taken for the linked online output (and that must exactly match the dvi output w.r.t line and page breaks, which it does nicely nearly all the time).

@barbarabeeton Why is it much easier?

BTW, is it possible to switch column layout midway in a longtable?

Though I'm not sure in that case if it could be called the same table...

@FaheemMitha -- Each graphic is distinctly separate, so it's easy to tell if (a) all fonts are embedded, (b) after scaling, the fonts are large enough to be readable, (c) after scaling, that linke are thick enough so that they don't disappear on the press, etc., etc.

@barbarabeeton I see. And those things are not possible from a PDF?

@DavidCarlisle -- I don't remember (and my notes aren't sufficiently detailed). Maybe Chris does. (And if you are in touch with him, remind him that he said he'd send me his email address.) Thanks for the link to mathml4 -- I may look in, after we've got the proceedings under control.

@FaheemMitha -- As one of my former colleagues once put it, "once a document is in pdf, everything is all hamburger." You can't tell whether a particular line came from a graphic (or which one) or from the text. So everything must be checked manually. People time is very expensive.

@FaheemMitha no but you can stop one and start another on the same page

@DavidCarlisle Yes, that's the other option.

@barbarabeeton nearly? When breaks didn't match, what typically caused it?

@HaraldHanche-Olsen links, probably

@barbarabeeton I see. Interesting. I didn't know that.

@DavidCarlisle whatsits playing havoc with spacing?

@HaraldHanche-Olsen pdftex has some primitve support for linebreaking while maintaining link annotation, doing the same with dvips is "interesting"

@DavidCarlisle ooh scare quotes

(channeling @PauloCereda)

@DavidCarlisle I am, in fact, properly scared. I seem to have witnessed some discussions on this sort of issue right here, in chat.

@HaraldHanche-Olsen -- Yes, @DavidCarlisle is correct. It's usually links. And whatsits as you surmise. If I remember correctly, usually this could be corrected by changing the location of labels, where index terms are entered, that sort of thing. (I think some small mods were made to some definitions in class files to make this more robust; but almost no class file changes have been put into the public released versions, much to my disgust. The bugs list is enormous.)

@barbarabeeton Thanks. I expect you probably do remember correctly.

@DavidCarlisle We've sorted it, just need Frank to suggest a hook

Does anyone use the listings-ext package and the script listings-ext.sh?