@daleif I think you forgot to double the # when copying the textcase fall back definitions into the \IfFileExists{textcase.sty}-Argument. See tex.stackexchange.com/questions/517378/…

@DavidCarlisle Yes, because it has to happen after unpacking

@JosephWright having fun with Twitter, I see. :)

@PauloCereda I'm having fun with something Frank sent ...

@JosephWright ooh

@JosephWright yes I knew that once, I was just too tired to figure it out again last night, anyway that will account for a lot of the failures as the new test2e was always setting OT1 even in the TU tests (not good:-) but first I'm having fun with something Frank sent ...

@DavidCarlisle ooh more fun

@PauloCereda no fun for you.

@DavidCarlisle oh

changelog.complete.org/archives/…

@DavidCarlisle Enjoying the 'homework'? ;)

@JosephWright people should use ascii 8+3 filenames:-)

@DavidCarlisle Well clearly ...

@JosephWright ooh

@DavidCarlisle ooh

@DavidCarlisle PROGRA~1

@DavidCarlisle Makes our 'fun' with UTF-8 & space look easy

@JosephWright more fun

@DavidCarlisle Think we look quite sensible in contrast ... issues fixed in days, actually working to support stuff, ...

@UlrikeFischer you're right, I'll have a look (never tested, oops). Though a simple doubling didn't seems to work in my test, I'll have another go

Seems to have been a manual error # -> ## works fine. I'll also add a warning to have users install textcase

@daleif @egreg @UlrikeFischer @JosephWright requiring ## in iffileexists looks like a 2e bug to me, but perhaps we can't fix it now, as changing the behaviour bound to break existing uses.

@DavidCarlisle is the behaviour old? (I don't remember to have checked or tried it before)

@UlrikeFischer 2016 is the oldest I have and it is same there

@UlrikeFischer As I do not use miktex I don't know how the packaging is handled. In memoir the memhfix.sty adjustments to hyperref is generated by mempatch.dtx which used to provide both mempatch.sty and memhfix.sty, I'm considering renaming the source file to memhfix.dtx any idea if that will cause package problems in miktex? The provided sty and cls files will be the same

@DavidCarlisle that should be ok then. I had to remove tl2018 from my laptop - not enough space ;-(. I have to go now, disaster happened: my cooking plate is broken, I need to get some fall back until it is repaired.

@daleif normally, if you announced that at the ctan upload Christian should pick it up, if a packaging error happens it should be easy to fix.

If needed I have TLs the way back to 2004 packed down on disk

@UlrikeFischer thanks, I'll add that to the TODO list for the next regular release.

memoir v3.7j shipped

@DavidCarlisle I tried with TL2012 and it's the same.

@egreg yes latexrelease.sty thinks it's always been that way (still wrong though...)

@DavidCarlisle Certain is in TL'09, which likely means 'for ever'

@JosephWright ooh

> upright italic shape (more a curiosity)

@DavidCarlisle wow, you are in a lot of organisations. :)

@JosephWright who said that?

@UlrikeFischer FMi

@UlrikeFischer Document Frank linked to last night, p48

@JosephWright oh, you are reading too. But I overlooked this.

@JosephWright But I found the "one has to love with it"

@UlrikeFischer Ah yes, I'd seen that; there's also a 'nd' in place of an 'and', but otherwise no typos so far

@UlrikeFischer It was set as homework ;)

@JosephWright ooh

@JosephWright I found a few more (adding sticky notes on acrobat as I go past:-)

@DavidCarlisle Cool

@PauloCereda you should give @JosephWright a uktug github avatar as the H there is annoying:-)

@DavidCarlisle ooh

@JosephWright you are not in the tug github yet?

@DavidCarlisle In all seriousness, I think the website wants redoing: I am thinking about looking for some 'real' design work ...

@DavidCarlisle No, I'm clearly not important enough ;)

@JosephWright WHAT

@PauloCereda Well, neither is Karl, apparently, which seems .. odd

Everyone got PR2?

@JosephWright Wait, does Karl exist outside mailing lists?!

@JosephWright Karl is (most people are set as private) there are 6 by now

@DavidCarlisle Ah, right, I did wonder about that

@JosephWright LaTeX2e <2020-02-01> pre-release-2 (development 2019-11-21 branch)

Happy New Year :)

@PauloCereda he exists at least in mailing lists, facebook, and in a C library for file searching.

@DavidCarlisle ooh

@PauloCereda if you get past Christmas dinner

@DavidCarlisle oh no

@DavidCarlisle Too late: it's a feature

@egreg agreed

Strange question for dtx files. Can one split a dtx file up into several smaller files, input these and still have docstrip generate the resulting cls or sty. My naïve tests with \input{t1} and % \input{t1} both seems to fail.

(reason: for latex dtx files emacs' doctex syntax highlighter is super slow)

does pgf plots support error bars in datetime plots?

@daleif Yes, but you only \(Doc)input them for the LaTeX run. For the docstrip run you need to point them out in the .ins file

@daleif Like this: \generate{\file{package.sty}{\from{source1.dtx}{package}\from{source2.dtx}{package}}}

@PhelypeOleinik right, that seem to work, thanks, l just have to remember to disable \Finale in the split state

@daleif \AlsoImplementation should do that, I think

@daleif Unless you don't want to typeset the implementation... Then I think you need to deactivate it manually

@PhelypeOleinik the manual is in a separate document, so it makes no sense here to omit the implementation. I do the memoir manual in a similar fashion. My work copy is split into several files so I can use \includeonly` and when I build the release a script merges everything into the single files that are the memoir manual src on CTAN.

@daleif Cool. The memoir manual is huge, so it only makes sense

@Skillmon twitter.com/AcademiaObscura/status/1197528057164288000

@PhelypeOleinik well on model computers it compiles pretty fast, but my home pc is not as fast as my work pc ...

@UlrikeFischer @JosephWright bunch of failures of the form

*** ../build/test/tlb-rollback-002.luatex.tlg	2019-11-21 14:58:00.044913200 +0000
--- ../build/test/tlb-rollback-002.luatex.log	2019-11-21 14:58:00.044913200 +0000
***************
*** 9,15 ****
  Document Class: article ....-..-.. v... Standard LaTeX document class
  (size10.clo
  File: size10.clo ....-..-.. v... Standard LaTeX file (size option)
! )
  \c@part=\count...
  \c@section=\count...
  \c@subsection=\count...
--- 9,15 ----
  Document Class: article ....-..-.. v... Standard LaTeX document class

@DavidCarlisle not my fault ;-)

@DavidCarlisle but perhaps we could add a option to make luaotfload less verbose.

... we could normalize that away in l3build (but may get line break diffs anyway) but more generally is there a way just to tell luaotfload not to be so noisy, as we are comparing with pdftex here it just never logging anything woul db egood

@UlrikeFischer much much much less verbose as in "just shut up" would be good:-)

@UlrikeFischer when does it give that font database message? most of the luaotfload messages come before \START but that one comes after class is loaded.

@UlrikeFischer @MarcelKrüger the log file for lualatex small2e which is the first thing we tell users to try is a bit well scary

(./small2e.tex
LaTeX2e <2019-10-01> patch level 3
Lua module: luaotfload-main 2019-11-10 3.11 luaotfload entry point
Lua module: luaotfload-init 2019-11-10 3.11 luaotfload submodule / initializatio
n
Lua module: lualibs 2019-11-06 2.69 ConTeXt Lua standard libraries.
Lua module: lualibs-extended 2019-11-06 2.69 ConTeXt Lua libraries -- extended c
ollection.
Lua module: luaotfload-log 2019-11-10 3.11 luaotfload submodule / logging
Lua module: luaotfload-parsers 2019-11-10 3.11 luaotfload submodule / filelist

@UlrikeFischer but for the purposes of the test suite it would be good if I could inject some Lua so it put OMIT TIMO around each of its log messages when I am trying to emulate a pdftex log.

I suppose I could disable texio.write completely...

@DavidCarlisle try with a luaotfload.conf with the content

[run]
log-level = -1

@DavidCarlisle well the "LuaModule" messages are from the luatexbase.provides_module function and are basically equivalent to the \ProvidesPackages messages from latex.

@UlrikeFischer yes sure but here I'm trying to hide the fact that extra code is running to get (in most cases) the same log as pdftex, and it never makes luaotfload messages:-) But actually I think some of these are where I haven't reverted enough macro definitions and it's going through an OTF font path when it ought to be using tfm fonts, so I'll check a bit more before disabling it, thanks

@DavidCarlisle did you try the log-level setting? This should at least get rid of the db messages.

@PauloCereda looks too small to make a decent meal.

@Skillmon oh no

@PauloCereda but those shiny ears look cute.

@Skillmon awww <3

@DavidCarlisle I got an answer about mtpro2 - perhaps something will happen in the end.

@UlrikeFischer thanks I got distracted, will get back to this

@DavidCarlisle ooh was it the cute bunny?

@PauloCereda Probably duck hunting.

@AlanMunn oh no

@DavidCarlisle ^^ YOU ARE EVEN TALLER THAN JOSEPH?!

Wow.

Just discovered you can clone the GitHub wiki for a repo

@JosephWright yes!

@PauloCereda github.com/latex3/latex2e/wiki/… :)

@PauloCereda Joseph is probably sitting and writing the minutes.

@UlrikeFischer ooh

@UlrikeFischer Er ...

@UlrikeFischer During the UK-TUG meeting, it was possible to hear keystrokes. He was writing the minutes!

@PauloCereda Er, yes, I did make a start

@JosephWright <3

@barbarabeeton one nice Czech publisher just published a book called Providence by Moore and Burrows. I wouldn't have known it unless a good friend and a graphic designer posted on facebook a link to a funny article about some tough and irreversible mistakes in the typesetting of the book.

Got me a new 24" monitor.

@PauloCereda OOOH

@JosephWright it should be a curved one, but I decided to be conservative. :)

Would you say a token of character code 127 and category code 4 followed by a token of character code 127 and category code 3 is more seldom than two tokens of some private "namespace" e.g. \mytok@A\mytok@B?

probot.github.io/apps/pull

@Skillmon Most of the time, I'd go for a private namespace as it's clearer (you can't see catcodes)

@JosephWright but what would you expect to be more robust in case you'd use this as argument delimiters?

@Skillmon Still private tokens: think about tracing stuff through

@JosephWright I never trace :) My code works \end{sarcasm}

@Skillmon The onyl time I use weird-catcodes is where I need them, usually in the 'one charcode, two catcodes' case where you need something the user can't generate 'accidentally'

PROBOT

@JosephWright ^^ :)

@JosephWright (case in question, the code tex.stackexchange.com/a/517265/117050 for "fast \ifempty) my use case is exactly that I want tokens that no-one can input accidentally, I used \lccode`\$=127\lccode`$=127 there as I think this should be more robust, but I wanted to hear others' opinions on this.

@PauloCereda You were worried that a large screen would be difficult to read with your adaptive eyeglasses. What is the end result? What made you change your mind?

@AndréC The price was inviting and it seems I can tune the screen size in my Linux box, so I am willing to give it a try. :)

lifehacker.com/…

@yo' -- I'm not familiar with this. I'll have to look it up.

@barbarabeeton I have no idea what it is, but it cought my attention.

@PauloCereda (I probably shouldn't mention the size of my monitor)

With a 7" screen it is easier to focus on the TeXing.

@yo' -- It appears to be (in graphic form) about H.P. Lovecraft, a Providence author of "fantastic" stories. What sort of dreadful errors were perpetrated?

@barbarabeeton layers smashed in a wrong way, so missing text in comics bubbles etc.

@yo' ooh

I just finished an interesting mini-project – getting a bunch of data out of a postgres database using luasql and feeding it into a longtable. No intermediate files or anything; it all happened in a single luatex run.

@HaraldHanche-Olsen yay!

@PauloCereda only when I am standing on his feet

@HaraldHanche-Olsen sounds great, but intermediate files have the advantage of saving time on future runs, so in most of my scripts I opt for intermediate files (though I never used SQL, so this is more in general)

@DavidCarlisle ooh

Got some books from CS Lewis

@Skillmon Nah, the postgres queries are so fast, time is not a problem.

But this is the first time I needed the longtable package. It works, oddly enough.

@PauloCereda that's the room we used this year as well

@DavidCarlisle :)

@DavidCarlisle Nicer than the Dungeon

@DavidCarlisle In former times people used to stand on the shoulders of giants.

@DavidCarlisle ooh

@JosephWright that's not saying a lot

@JosephWright bah :)

@DavidCarlisle :)

@yo' -- oops! Well, the English version is here in the library, but not the most local branch. I'll take a look when we get back from our trip. (Away next week in Quebec City. Will still have computer connection, but ordering up the book won't get it here until after we leave.) And will report.

@DavidCarlisle -- I think I've met most of the people in the picture, but I know the names of only half of them. So Chris decided to brave the November weather and come back to London for the meeting.

@barbarabeeton that was a a couple (or 3?) years ago, not this year

@DavidCarlisle At least three years I think

@DavidCarlisle uk.tug.org/2015/11/01/speaker-meeting-2015-2

@JosephWright somehow align breaks my <DEL> tokens with strange categories :(

@Skillmon you are not doing catcode changes in the align are you?

@Skillmon oh you used the alignment category, bad choice:-)

@DavidCarlisle no, I'm using a macro which is defined to use <DEL> tokens of two category codes as delimiters, and one of the categories is a alignment tab, so I'm stupid.

@Skillmon Oopsy

@DavidCarlisle it just occurred to me... Will change the choice.

@JosephWright was just stupid to use that category, I guess I should use letter and other, those should be pretty safe.

@PauloCereda This photo needs a caption.

Well, the page it originates in does have a caption. Though only one caption for two photos.

Actually, I have a feeling I've seen this page before.

@UlrikeFischer I see LuaHBTeX For The Win: tex.stackexchange.com/questions/517460/…

@JosephWright ;-) artful method to get people to use (lua)latex-dev ;-)

@UlrikeFischer :)

@UlrikeFischer is the plan to make harfbuzz the default?

@JosephWright it feels a bit odd to see the future here while in another place there is a long discussion how to get back to 2015 ;-).

@UlrikeFischer :)

@DavidCarlisle there is no plan yet. But I think we should be careful here. If people used some of the handlers to adjust kerning or so it could all break - I haven't yet tried how this combines with harf mode. Will could perhaps make it the default for a number of scripts.

@UlrikeFischer, From you comment here, how current a TeXLive is needed for this? It doesn't yet work on Debian testing which has a version from 20191112.

@DavidPurton you need luaotfload 3.11 and as engine luahbtex. Texlive mapped lualatex-dev to this engine around a week ago.

@DavidPurton if you have the engine you can look if you have a luahblatex.

@UlrikeFischer Debian doesn't seem to be shipping luahblatex yet at all!

@DavidPurton can you use tlcontrib?

@UlrikeFischer I haven't tried trying to use new binaries from tlcontrib alongside the Debian packaged ones. I don't really need these features at the moment. I'm just hoping that it will get into the main Debian packages soon

They do have luaotfload 3.11 though

If TeXLive changed things around a week ago, perhaps I'll see Debian bring it in next month

@DavidPurton that is a bit older (we needed to be prepared for the new engine ;-))

Though, the Debian version of the binaries is a fair bit older (18 Sep)

@DavidPurton note there will not be a format called luahblatex (although some older test releases had that name) it will just be called lulatex (in 2020) or since this week in texlive 2019, lualatex-dev

@DavidCarlisle Don't I still need a luahbtex binary though?

That will link to libharfbuzz?

@DavidPurton yes, (which got added last week) but I meant that if you were looking out for a luahblatex to appear in the debian version, that won't happen. It may not happen in debian either, luahbtex is in the texlive sources for 2020, it isn't fully back ported to texlive 2019 release they just added it (for the main three platforms) more or less by hand, copied from tlcontrib) So It may not reach linux packaged versions until tl2020 (I'm not sure)

@DavidCarlisle OK. Sometimes Debian does back port things. We'll see what happens.

waves

If anyone uses the JetBrain's TeXiFy IDEA plugin by Hannah-Sten, and writes in a character set that features accents, or non ascii characters, do you experience strange highlighting upon said accents and characters? all my "é" get show in red, and I'm unsure why. Should TeX files always only be written in ascii characters?

please note, that everything works fine, the output result is ok, the plugin is only hightlighting the letters, and I'm wondering if I am missing some configuration to tell the document that it's to be written in utf8

@FélixGagnon-Grenier by tex files do you mean latex? latex assumes utf-8 by default (in current releases) so certainly don't have to be ascii

hmmm, yes, I do mean latex, however I thought "tex" was the larger family

oh, you mean that other ones would have different behaviour

@FélixGagnon-Grenier sure but the utf8 support is latex code, not plain tex

right, that must be why it works

oh and that could also be why the plugin presumes not of my specific latex but of the larger tex family

@FélixGagnon-Grenier given that the overwhelming majority of tex is latex, it's worth raising with them (but as you say you may need to tell the editor the file is utf8, I don't know it at all)

yes, I was wondering if it was a worthwhile option, I think I will, thanks :)