@DavidCarlisle It has something to do with my use of TikZ's externalisation library. Either \tikzsetfigurename or \tikzsetnextfilename. But every time I think I've pinpointed the problem it vanishes in a puff of nothing.

@DavidCarlisle I can make this more minimal, but this is the minimalist I'm going to get right now, I think.

\begin{filecontents*}{\jobname-img.tex}
\csname standaloneignore\endcsname
\documentclass[tikz,crop=false,preview=false]{standalone}
\begin{document}
\begin{tikzpicture}[baseline=(current page.north west)]
  \shade [top color=black, bottom color=black, middle color=white] (current page.north west) rectangle (current page.south east);
\end{tikzpicture}
\end{document}
\end{filecontents*}
\begin{filecontents*}{tikzlibraryprawf.code.tex}
\ProvidesFile{tikzlibraryprawf.code.tex}
% RHYBUDD: regex yn l3 experimental

\begin{filecontents*}{prawf-img.tex}
\csname standaloneignore\endcsname
\documentclass[tikz,crop=false,preview=false]{standalone}
\begin{document}
\begin{tikzpicture}[baseline=(current page.north west)]
  \shade [top color=black, bottom color=black, middle color=white] (current page.north west) rectangle (current page.south east);
\end{tikzpicture}
\end{document}
\end{filecontents*}
\begin{filecontents*}{tikzlibraryprawf.code.tex}
\ProvidesFile{tikzlibraryprawf.code.tex}
% RHYBUDD: regex yn l3 experimental

@DavidCarlisle It is half the length it looks. I edited the first line and the bloody system has posted the whole thing again, rather than correcting the first version. The first version won't work unless you save your test code to prawf.tex, which seems, on the whole, somewhat unlikely.

@cfr Same here

@cfr I'll sort it

Quack!

@cfr @JosephWright Hello there! What strike?

@PauloCereda Qua qua qua!

Hi all

@Buochserhorn Aloha!

I've been using LaTeX for about two years now and have always used \setlength\parindent{0pt}, as these indents really bothered me at the beginning. Now, as I'm writing longer texts, I'm kinda warming up to them and considering that Knuth is somewhat a god of typography I'm thinking about "introducing indents to my life". What are your thoughts on indents? Do you like them? Do you always use them? For what reasons?

@cfr will look later thanks

@Buochserhorn In my case, we have to follow the guidelines of our technical standards association.

@Buochserhorn you need some way to mark a paragraph, either horizontal indent space or vertical parskip space (but not both)

@Buochserhorn I always use indents in longer pieces. I sometimes use no indent, but positive parskip, in shorter items. It sometimes seems to work better in handouts of a page or two, and the like.

Thanks a lot for your inputs!

@Buochserhorn I think it's fancy the \noindent at the start of a chapter and subsequent indentations. :)

@Buochserhorn Right now, I am putting together a conference proceedings volume. Paragraph indent rules, in this case. Many contributions have long stretches of displayed equations with a line or two of text in between. In this setting, paragraph indents look weird, but at least you can see them, as opposed to parskips. And I often have to guess whether a new paragraph after a displayed formula was intended or not. Many authors seem to have a rather relaxed attitude about it.

@PauloCereda Absolutely. But that should be automated whenever possible.

@HaraldHanche-Olsen Exactly. :)

@PauloCereda \noindent isn't used in the implementation of that:-)

@DavidCarlisle Thought so. :)

@yo' ucu.org.uk/strikeforuss

@JosephWright We're not really big enough to have any voice, but are affected by the outcome...

@JosephWright ah it's inside academia world, that's why there are no headlines on BBC :-)

looks quite serious to me, but of course I can't know anything from a 30-second glance.

@DavidCarlisle Ah, of course

@yo' You might see it on Thursday: the great irony here is that we have a week of no teaching, so it's a bit awkward

@JosephWright wow

@JosephWright you could stop working on l3draw for the day

@JosephWright some "timing", one would say :-)

@yo' Oh, my uni has had the day set up for ages (it's an idea from the top to have 'Do Something Different' week for the students): just one of those things

@yo' The union have called 15 strike days ...

@JosephWright well, the early schedule here is quite bizarre to, and different between universities (even in the same city), sometimes even between faculties

@JosephWright that's a pretty long strike

(btw, is "union" a "syndicate"? Just trying to understand what's going on)

@yo' That I don't follow: UCU is the union for academic and academic-related staff in universities founded before 1992

@yo' "union" == "Trade Union" organisations representing rights of employees

@JosephWright @DavidCarlisle dammit, that must be the French influence. What you call a union is un syndicat in French. Sorry for the confusion

@JosephWright And now for something completely different.

any miktex users (or babel 3.18 users) around?

Sometimes you don't get bug reports: I've used your LaTeX MSU thesis class for my dissertation, and it simply worked perfectly. The graduate school had just approved my thesis without any formatting issue. Your package has literally saved a lot of trouble for me compared with the experience from my other folks who are using MS word.

@AlanMunn Yay!

@AlanMunn then you wake up from your dream?

@DavidCarlisle (a) arara does not work with Cygwin. (b) Paulo makes it compatible. (c) Now arara does work with Cygwin. (d) The English bloke does not use arara. :D

@PauloCereda I would use it if you re-wrote it in emacs lisp

@DavidCarlisle lol

@DavidCarlisle ooh :)

@DavidCarlisle Another blank chapter delayed.

@PauloCereda do not start re-implementing arara until you have finished your thesis

@AlanMunn looks like we both had same thought

@DavidCarlisle ack

@DavidCarlisle @PauloCereda do not start re-implementing your thesis until you have finished your thesis

@AlanMunn uhhh

@PauloCereda Trust me, this is very wise advice.

This is getting too complex.

@AlanMunn My thesis uses Lua. :)

@PauloCereda Do you howl at it?

@AlanMunn well played sir. :)

@PauloCereda You could also say "My thesis uses C++" or "My thesis uses Atlas autocode" with equal validity

@DavidCarlisle That's true. :)

@AlanMunn -- congrats! aren't thank-you notes nice?

@barbarabeeton Yes, very. And even the bug reports usually come with some appreciation too, just with an added 'but'. :)

@AlanMunn -- perhaps you should be conscripted, er, encouraged to write a little treatise on how to write a proper package that is worthy of such respect. you must follow a few useful guidelines.

@barbarabeeton Quack!

@PauloCereda -- quack to you too.

@barbarabeeton also, Fibonacci. :D

@barbarabeeton Well much of the credit goes to memoir. And testing, testing, testing. With a good test document.

@AlanMunn -- ah. it sounds like you follow my principle of using "good bad examples". find the most peculiar case that has actually happened, and is necessary and/or justifiable, and make sure that it is satisfied by your implementation. (nothing is foolproof, and fools are becoming more devious.)

^^^ i was given this duane bibby drawing as a reward for having survived 55 years at the ams. it was a present from don knuth.

@barbarabeeton Beautiful!

And I want to point out that I saw the Fibonacci sequence before everybody!

@PauloCereda some people had a head start

@DavidCarlisle bah what do they know

@PauloCereda personally I was most pleased to finally see BARBARA in capital letters

@DavidCarlisle LOL

@barbarabeeton I remember when my supervisor celebrated an important jubilee (55) and said that she has to very enjoy herself as the next one is quite far away...

Btw, is there a word for people who are overthinking their overthinking? :-)

@yo' overᵒᵛᵉʳᵗʰᶦⁿᵏ

:)

@PauloCereda :)

@yo' I think that requires stackengine.

@AlanMunn or stackoverflow

There seems to be some clash between the listings package and unicode-math? MWE ↓↓

\errorcontextlines99
\documentclass{article}
\usepackage{listings}
\RequirePackage{unicode-math}
\lstset{basicstyle=\small}
\begin{document}
\begin{lstlisting}
  % For-loop
\end{lstlisting}
\end{document}

Does this ↑↑ compile for any of you, in lualatex?

! Missing number, treated as zero.
<to be read again>
global
lst@whitespacefalse ->global
                             let lst@ifwhitespace iffalse
lst@ProcessDigit ->lst@whitespacefalse
                                       lst@ifletter expandafter lst@AppendLe...

7->lst@ProcessDigit
                    7
l.1 7
   0
lua_now_x:n #1->luatex_directlua:D {#1}

__um_fontdimen_from_param:nn ...t.current(),"#2")}
                                                  scan_stop:
__um_luatex_copy_fontdimens: ...tPercentScaleDown}

@HaraldHanche-Olsen minus signs are tricky

@DavidCarlisle People always tell me to be positive. You too?

Registration for TUG2018 is open: tug.org/tug2018/register.html

@DavidCarlisle I am a bit puzzled by <argument> $-, though. Wonder where it came from.

@DavidCarlisle Oh, wait, I guess it is a listings thing to actually typeset minus signs in mathmode?

@DavidCarlisle If I insert {\small$-$} before \begin{lstlisting}, it compiles.

@DavidCarlisle So the listings code is trying to process part of the expansion of \check@mathfonts, and things go pear shaped ever after? Not very robust, if you ask me.

@HaraldHanche-Olsen As a workaround: \lstset{basicstyle=\small,literate={-}{{{\symbol{`\-}}}}1}

@egreg Well, braces are entertaining. What's the 1 doing in there?

@HaraldHanche-Olsen The problem seems to be in the fact that listings tries to change hyphens into minus signs. Possibly less braces suffice.

@HaraldHanche-Olsen One pair less is sufficient; the 1 means how many slots the literate character is supposed to occupy.

@egreg Ah, okay. (I never used the listings package myself; I am fighting someone else's code here.)

@HaraldHanche-Olsen If you want to use a minus sign, use \symbol{"2212} instead of \symbol{`\-}

@egreg Yes, that works, also in the original document.

@egreg Only remaining question, then: whom to blame?

@HaraldHanche-Olsen The disappeared author/maintainer of listings, of course.

@HaraldHanche-Olsen Great piece of software, but the combination between minted and pygmentize is better.

@egreg Blaming disappeared persons rarely helps.

@egreg I'll keep that in mind. This particular paper is full of matlab code. I wonder how much work it would be to change to those packages? If I hit no more snags, I may prefer just to stay with listings, though.

@egreg But thanks for the workaround, anyhow.

@HaraldHanche-Olsen so blame @WillRobertson seems a reasonable alternative in the circumstances.

@JosephWright Thanks - much appreciated.

@JosephWright :(

@yo' UCU has called for a month of industrial action. This is the union representing academics in higher education institutions in the UK. (Maybe further education, too.) Lecturers, researchers and academic-related support staff. 14 days of strike action over 4 weeks, with work-to-contract (action short of strike) on the non-strike days.This is the most serious industrial action in the university sector in recent times.

@cfr sounds serious. As I say, our retirement plan system is completely different from yours, so it's tough for me to judge the situation, but it certainly looks serious. I wish you best!

@DavidCarlisle I have a feeling it should be obvious. I can make it go away by commenting one line or another in the TikZ library, but I can't pin it down to anything specific. (That is, commenting different bits makes it go away, but adjusting those bits in the ways I think might help don't.)

@cfr Our hearts are with you!

@PauloCereda @yo' Thanks.

@cfr will look this evening, haven't had any time today

@AlanMunn Ah, but possibly that predates the introduction of the literate key?

@HaraldHanche-Olsen Harald, are you a member of a TUG group?

@HaraldHanche-Olsen I don't think so.

@PauloCereda No, just a plain member of TUG.

@HaraldHanche-Olsen Ah thanks. :)

@AlanMunn I only asked because the answer seems to do the same as using literate does. But perhaps I have missed the point.

@HaraldHanche-Olsen I'm not sure then. I use listings only occasionally.

@DavidCarlisle No rush. I wasn't nudging you. More thinking aloud.

@JosephWright I'm also having issues with verbatim content in the appendix, even with the replacement .sty. For example, I have

\begin{verbatim}
  \newcommand\tableofcontents{%
    \section*{\contentsname
      \@mkboth{%
        \MakeUppercase\contentsname}{\MakeUppercase\contentsname}}%
    \@starttoc{toc}%
  }
\end{verbatim}

This is fine when I compile the handouts or notes. But when I compile the slides, I get an error complaining about an undefined control sequence, even though this is verbatim and ignorenonframetext is active.

Package hyperref Warning: Difference (2) between bookmark levels is greater
(hyperref)                than one, level fixed on input line 535.

! Undefined control sequence.
\MakeUppercase ...ppercaseUnsupportedInPdfStrings

l.561 ...ntentsname}{\MakeUppercase\contentsname}}
                                                  %
?
! Undefined control sequence.
\MakeUppercase ...ppercaseUnsupportedInPdfStrings

l.561 ...ntentsname}{\MakeUppercase\contentsname}}
                                                  %

@cfr ^^ cheer up

@JosephWright This could be a feature request for siunitx

@PauloCereda ;) Did you get your post yet?

@JosephWright The problem is not hard to avoid. But something has definitely broken code which did work.

@PauloCereda Thank you!!

@cfr oh not yet... But I have faith! <3 I got a Merry Christmas from Nicola Talbot last Wednesday! :P

@egreg Hmm, yes: funny it's not come up before!

@PauloCereda: You've got mail ...

@ChristianHupfer you too. :)

@PauloCereda Oh my :D

@marmot wow!

@PauloCereda Aha ... getting closer, then?

@cfr I believe so!

@PauloCereda: We should sent Christmas Greetings to you in June already ;-)

@ChristianHupfer bring them to TUG :D

@yo' I am not attending

@cfr ......\glue 3.33333 plus 1.66498 minus 1.11221

@cfr ^^^ vvvv

\documentclass[tikz,crop=false,preview=false]{standalone}
\begin{document}%

What's with all the linguex questions lately ?!

@Skillmon probably @AlanMunn setting exam questions

does anyone understand microtype (@UlrikeFischer:-)

@JosephWright being on strike, he probably can't answer this one :) tex.stackexchange.com/q/416283/11002

@yo' I'll see what I can do: need to ship l3draw ...

@DavidCarlisle I'll get ALWAYS SECOND engraved in my tombstone

@DavidCarlisle better ask Robert ;-) But I can offer this example:

\documentclass{article}
%\usepackage{fontspec}
\usepackage{microtype}

\begin{document}
\scshape a
\end{document}

@UlrikeFischer hmm 815 is the glyphindex of /two.oldstyle but I can't see why just that one is going negative

@DavidCarlisle There a lot negative numbers. The problem seems to be the "\MT@info@missing@char" message.

@UlrikeFischer see, i knew you were the person to ping:-)

Package microtype Info: Character `/two.oldstyle' has a width of 0pt
(microtype)             in font `TU/lmr/m/sc/10'.
(microtype)             Ignoring protrusion settings for this character.

@DavidCarlisle ;-) It basically does \iffontchar\MT@font\MT@char@ and this breaks for negative numbers. It should probably be \iffontchar\MT@font-\MT@char@, but this is to Robert to decide. Who should notify him?

@UlrikeFischer you're on first name terms:-)

@UlrikeFischer yes I just commented out the lines in the above

\def\MT@info@missing@char{%
  \MT@info@nl{Character `\the\MT@toks'
%^^X    \iffontchar\MT@font\MT@char@
      has a width of 0pt
%^^X    \else is missing\fi

@UlrikeFischer but I'm not sure it's supposed to be missing, as far as I can see /two.oldstyle is in the font and XeTeXglyphindex returns 815, so why does MT@char end up being -815 and an illegal char number?

@UlrikeFischer in particular why does microtype-xetex.def do \edef\MT@char@{-\the\@tempcnta}%

@DavidCarlisle I really hate linguex. :)

@DavidCarlisle for some reason /two.oldstyle seems to have width 0 when fontspec is not loaded and so is reported missing. The number 815 seems to be wrong, it should be 808.

@UlrikeFischer yes I noticed your test acted differently when fontspec is loaded, which is odd, as it's using the same font

@DavidCarlisle The problem is the \ifMT@fontspec\MT@font\fi in the setup. when fontspec is not loaded microtype doesn't correctly activate the scshape and so actually retrieves the glyph index from the roman font:

\documentclass{article}
%\usepackage{fontspec}
%\usepackage{microtype}

\begin{document}

\the\XeTeXglyphindex"two.oldstyle"
\scshape
\the\XeTeXglyphindex"two.oldstyle"

\end{document}

I noticed in a trace MT@font ended up with different definition but don't really know the code well enough to see which diffs are important:-)

so you should probably post an answer saying load fontspec (as a better workaround than mine) @UlrikeFischer

@DavidCarlisle I did it, I also wrote Robert.

@UlrikeFischer thanks

@DavidCarlisle Oh, gods. Thank you. I thought I'd put % everywhere (regardless of whether it was in expl3 space or not, I still added them. But not there. Though I'm even more confused about why commenting out certain lines made the space disappear, if that's the problem. Thnak you.

@cfr never really looked at the standalone doc, so perhaps crop=true would have been better than crop=false then it may have removed the space on its own? that said it seems like standalone ought to start with \ignorespaces but apparently not