@DavidCarlisle No, just seen they added it.

@egreg oh yes, I suppose we need to decide if tl2018 should build a lualatex53 by default (@JosephWright)

@DavidCarlisle That's where the fun begins with the tests. ;-)

@egreg we ran the test suite a while back I seem to remember it didn't show any issues (but it doesn't test that much lua code of course)

Could someone with a gold fonts badge please close this question

as an exact duplicate of this question

?

Thanks!

@HenriMenke I have no such badge, but I cast a vote anyhow. If one more vote comes in, there will be a tie for the duplicate. (I don't know what happens then.)

@HenriMenke Ah. It listed all three.

Palindrome ;-)

Is it better to use fake small caps or to include true small caps from a foreign font? Or is both a terrible typographic sin?

@JohnDorian Both are okay, so long as you do penance by working exclusively in vim for a year.

@JohnDorian (But I would go for the foreign font, if one can be found that goes well with your other font(s).)

@JohnDorian I have one of those too:-)

@DavidCarlisle: linux.slashdot.org/story/18/03/17/0438256/…

@DavidCarlisle ooh

@egreg a space is too much \these ction. Thanks, upvoted.

@PauloCereda I've lost faith in democracy, so I'm not surprised

@DavidCarlisle :)

@PauloCereda Fake news: they also voted Debian as the best distribution.

@egreg :)

Will pgf-ornaments be contained in TeXlive 2018?

@JohnDorian Huh?

@JohnDorian Is it on CTAN?

@JohnDorian And in TL

@DavidCarlisle Hans was pretty clear that wasn't what was expected/desirable

@PauloCereda Oh lol you're right

Just another stupid question by me then :-)

I don't think that question adds any value to this site. It is a localozed problem nobody can reproduce and the self-answer is just as cryptic.

@Johannes_B I've closed as unclear

@JosephWright just driven through Northampton to Market Harborough, turned round and drove back:-)

@DavidCarlisle Didn't like Market Harborough?

@JosephWright they decided "rather late" to cancel due to snow, somehow expect everyone checks their email while driving:-) There wasn't that much snow anyway

@DavidCarlisle Ah right

@DavidCarlisle Certainly snow here is not an issue

With so little reputation I get special numbers all the time :-)

@JohnDorian but also you found a way to get from 1 mod 5 to 3 mod 5, I tend to get stuck at a single value mod 5 (unless someone evil like @AlanMunn downvotes me of course)

@DavidCarlisle Who me? Never. :)

@AlanMunn I finally managed to acquire enough downvotes to get back to a 4*4 cycle :-)

@DavidCarlisle Well you have to accept lots of answers if you want to be a mod 2 guy :-)

@JohnDorian I have never accepted any answers (which is why it is difficult to change the rep mod 5 as upvotes give 10 and green ticks give 15. So the only other option is someone downvoting

@DavidCarlisle See? I think that's your problem.

@DavidCarlisle Alan is not evil, he's mean

I will make pizza tonight ...

@UlrikeFischer Mmmm pineapple.

@DavidCarlisle Muito obrigado.

@DavidCarlisle oh

@AlanMunn For confirming that you are evil rather than mean?

@UlrikeFischer got the pineapple ready?

@DavidCarlisle as chess players say "the thread is often stronger than the execution"

@DavidCarlisle :)

I wonder how long it will take @egreg to post an \ooalign answer to the math accent question?

@DavidCarlisle \ialign seems better.

@DavidCarlisle Quick questions about your edit tex.stackexchange.com/revisions/419999/3 I totally agree that the {texstudio} tag is unrelated to the questions, but is there any specific reason you removed the {package} tag? From the description it sounded like a reasonable good tag, compared to {beamer} which looks rather unrelated, as metropolis is a separate package and not part of beamer.

@samcarter The packages tag is unfortunately one of those catch-all tags that isn't terribly useful, I think, since almost every question involves a package in some way. So it seems to be better kept for questions about "what package does X" or "how do I install package X", which this question isn't about.

@UlrikeFischer Have you hear about Cracco's pizza? We Italians are able to quarrel about pizza also among us!

@samcarter @DavidCarlisle On the other hand, I probably wouldn't have removed the texstudio tag, since although the editor isn't directly responsible for the error, the setup of the editor presumably is, since it and TeXShop produce different results for the OP.

@samcarter I must admit I didn't think about it too hard but as far as latex is concerned metropolis isn't a package it's a beamer theme, so packages seemed misleading

@CarLaTeX After seeing that, pineapple pizza must look better and better to you. :P

Nutella pizza is good.

@AlanMunn I agree, and much cheaper!

@AlanMunn that one i did wonder about, but I had a level of scepticism in the OPs description (for example are texshop and texstudio on the same machine...:-0

@CarLaTeX This comment will haunt you forever. :)

@AlanMunn LOL

@DavidCarlisle Well, there is that possibility. But it's also possible that TeXShop finds MacTeX and TeXStudio finds an old MacPorts version.

@CarLaTeX no what it is about? I find only italian sites ..

@UlrikeFischer Here's an English version: washingtonpost.com/news/food/wp/2018/03/13/…

@DavidCarlisle But isn't a beamer theme just a package? I mean it is a .sty file that starts with \ProvidesPackage....?

@samcarter Technically yes, but functionally, no. It makes sense from the user point of view to treat packages as those invoked with \usepackage.

@PauloCereda Nutella pizza is a focaccia :P

@samcarter no it's a class it's a .cls file that starts \ProvideClass

@DavidCarlisle Not the themes...

@AlanMunn sorry I misread @samcarter's comment I thought said beamer was a package

@DavidCarlisle (unrelated) Can a class file encoding affect a document? I have a tiny class that I'm writing, but when I add \usepackage[utf8]{inputenc} to a document is gives me "Keyboard character undefined in enoding UTF8". But if I don't load my class, that error goes away.

@DavidCarlisle Ha, it's the evil gb4e.

@AlanMunn yes

@AlanMunn What have the Unicode Consortium ever done for us?

@PauloCereda In this case it's "What have active characters ever done for us?"

@AlanMunn :)

@DavidCarlisle I said that \ialign is better.

@egreg 56 minutes! slacker

@DavidCarlisle I had other things to do. The problem is not very interesting.

@egreg sure that's why I left it for you:-)

@DavidCarlisle, @egreg Get on with LaTeX3 then!

@JosephWright Generic accents could be a welcome addition to unicode-math.

@egreg A job for Will!

@egreg I'm thinking about \__ functions mainly, also drawing support

... and l3build, and ...

@DavidCarlisle How do I reset counters properly with tabulary?

\documentclass{article}
\usepackage{array,tabulary}
\newcounter{rownum}
\newcolumntype{R}{>{\stepcounter{rownum}\therownum.\quad}l}
\begin{document}
\begin{tabulary}{\linewidth}{RL}
foo & bar\\
foo & bar
\end{tabulary}

\end{document}

@AlanMunn they are supposed to reset themselves (if they are latex counters) if

@AlanMunn let me try...

@AlanMunn oh tabularx resets the counter but tabulary ...

@DavidCarlisle So this answer from 5 years ago never made it into the code, I guess. tex.stackexchange.com/a/119575/2693

@AlanMunn no but my answer now started with the same thought as then, so at least I am consistent

@DavidCarlisle :)

@AlanMunn An hour is long for me, five years is too short for @DavidCarlisle

@AlanMunn can you hear a strange grumbling noise in the background?

@egreg I can post the MWE as a new question, and then you can answer it first if you like.

@DavidCarlisle For the moment, pasting that redefinition into my class does the trick. But it would be nice if it were incorporated into the tabulary code.

@AlanMunn yes I should at least put a link to that answer as a github issue on the source so it doesn't get lost. will do that later.

@DavidCarlisle While I'm at it, is is possible to wrap a tabulary in a custom environment?

@DavidCarlisle Doh. Should have read the documentation for tabularx. :)

@AlanMunn generally that's a good idea although apparently not all of tabularx is copied to tabulary

Anyone know how to use siunitx with variables in place of numbers? For example, \SI{\pi}{m} doesn't seem to work. Am I just doing something stupid?

@AlanMunn github.com/davidcarlisle/dpctex/issues/12

@LoopSpace doesn't it have a don't parse numbers option thing?

@LoopSpace @DavidCarlisle is right, you need a parse-numbers=false.

@TeXnician that's the one, thanks:-)

@DavidCarlisle Ah, that's what that does. Thanks!

@TeXnician Thanks!

@AlanMunn you mentioned gb43 and utf8....

@DavidCarlisle Thanks. Fixed.

@DavidCarlisle Although I really need to bug Alexis to fix this.

@AlanMunn when was "a few years ago" what did we change?

@AlanMunn Slightly misread that ... I was imagining you saying "Alexa, fix all the bugs in my packages"

@AlanMunn I thought dental marks were like this ^

@LoopSpace That's something @DavidCarlisle would appreciate.

@AlanMunn yes I just realised that was the change

@DavidCarlisle I've deleted the answer temporarily, since there's a separate issue that my solution doesn't fix.

@AlanMunn I thought the problem is that both packages define \|.

@UlrikeFischer It's a bit of both, I think. But yes, gb4e does define \| (which I suspect nobody knows or uses...)

@AlanMunn Perhaps @PauloCereda should try "Alexa, write my thesis"

Hmm, @DavidCarlisle the parse-numbers=false is conflicting with fontspec. Specifically, if I \setmainfont{...} something then \SI{\pi}{m} no longer works. Any ideas?

@LoopSpace Show an example ...

@LoopSpace Now I understand why my test didn't work; I'm always using luatex.

@egreg -- can you look at tex.stackexchange.com/q/421766 please? i think the answer is wrong, although it works. i don't understand why the theorem heading is coming out italic, but i can't test it because we don't have any suitable font.

@barbarabeeton Curious.

@barbarabeeton It does the same with every font with Cyrillic support.

@CarLaTeX Quack

@barbarabeeton Investigating.

@UlrikeFischer Okay, it seems to depend on the order of \setmainfont{...} and \usepackage{unicode-math}. Here's two examples:

\documentclass{article}

\usepackage{siunitx}
\usepackage{fontspec}
\setmainfont{TeX Gyre Pagella}
\usepackage{unicode-math}
\setmathfont{TeX Gyre Pagella Math}

\begin{document}
\SI{\pi}{s}
\end{document}

That one works, but swap the \setmainfont and \usepackage{unicode-math} lines and it doesn't:

\documentclass{article}

\usepackage{siunitx}
\usepackage{fontspec}
\usepackage{unicode-math}
\setmainfont{TeX Gyre Pagella}
\setmathfont{TeX Gyre Pagella Math}

\begin{document}
\SI{\pi}{s}
\end{document}

@egreg -- thanks. let me know if it's a bug in amsthm that needs to be investigated; if it is, i should add it to the list.

@LoopSpace Neither work for me. (pi is missing, right?)

@LoopSpace I get no \pi with either example. siunitx does the equivalent of \mathrm{\pi} and this gives no output, but \SI[parse-numbers=false]{\mbox{$\symrm{\pi}$}}{s} works.

@marmot Wow, gorgeous!

@CarLaTeX I was pinging you because I thought you may have an answer... but not today, it's pretty late in Italy ;-)

@marmot No, it is not too late, it's only 9:40 p.m. :)

@CarLaTeX ^^^ Pomodori, mozzarella, carciofi, olive, cipolle, bacon, alici

@UlrikeFischer Did you put the cipolle in in order to scare the marmots away?

@UlrikeFischer Yay! Good, maybe only the bacon is strange :):):)

@CarLaTeX only because we didn't know the italian word. Is it speck?

Folks, anyone got any good pointers on why a caption being too long or too short might cause latex to crash?

either with a Underfull \vbox (badness xxxxx) has occurred while \output is active error

or at ("C:\path\to\document.bbl" [5])

@UlrikeFischer Speck is OK, but I think bacon is pancetta, which is also OK but I think we don't put them both with alici...

@barbarabeeton It's a problem with polyglossia, similar to one seen a few days ago: it has to do with \normalfont not acting on the actual main font.

@egreg -- thanks! (sure am glad it's not amsthm.) could you add a comment to that effect on the question, please.

@egreg What do you think about @UlrikeFischer's pizza: chat.stackexchange.com/transcript/message/43478051#43478051?

@UlrikeFischer With lualatex, and \setmainfont before \usepackage{unicode-math} then I get pi s. With the other way around, pi disappears.

@AlanMunn ^^^^

@LoopSpace Which unicode-math version do you have?

@UlrikeFischer 0.8i

@LoopSpace blame @JosephWright ?

@LoopSpace I have {2018/02/02} {v0.8l}.

@EmilioPisanty a warning message issued by the output routine isn't what I would call a "crash"

@DavidCarlisle no, as in, if I run it from terminal, -interaction=nonstopmode or not, it outputs one of those two messages, and then ceases to do anything at all

the only way I can get it out of there is Ctr+C

which then aborts the compilation

enter doesn't do anything, and neither does X or X+enter

@EmilioPisanty in which case it is presumably looping unless you mean you were at the * prompt.

@DavidCarlisle hmmmmm. I'm not sure what the * prompt is, but I imagine I'm not at it

(given that I don't know what it is)

how can I tell whether it's looping?

@EmilioPisanty this:

@EmilioPisanty you could add \tracingall just before the error and stop it after a while with control-c then the log will show what it was doing

as in, you've told it to \relax, and it's gone to one *?

@DavidCarlisle ok, lemme try that

@UlrikeFischer Mines 2017/11/18 v0.8i. I'm wary of updating since swapping the order is working for me, but isn't for you so I don't want to break it!

@EmilioPisanty yes but if you started from a file you would only have one * the two ** form is a special case where you don't give tex a file at all.

I'm running from terminal pdflatex -output-directory=build Manuscript.tex

@CarLaTeX Bacon? ;-)

@EmilioPisanty I have so little sympathy with anyone who uses -output-directory it makes things so much more complicated for no actual gain, but that's unlikely to be the case of the error here :-)

@barbarabeeton Not found the real problem, but I think it's a bug in polyglossia. Added a better workaround than defining a new theorem style.

@DavidCarlisle It's not affecting the error, but my directory structure is already complicated enough as it is ;-).

@EmilioPisanty so if you run that at the commandline do you get to a prompt or does it loop? (tex never stops so it must do one or the other of those)

@egreg We have the same feeling, then :):):) (@UlrikeFischer)

@CarLaTeX replace the bacon with nutella, then @PauloCereda will be happy

@CarLaTeX @egreg You Italians are never pleased.

@egreg @CarLaTeX should I replace it with pineapples?

@AlanMunn that's because we work so hard to annoy them

@DavidCarlisle if I run that at command line, I get this

@UlrikeFischer Any wurst could go

and it doesn't move from there

and the same thing minus the natbib warning if I run on the aux file of a previous successful compilation (with a much longer caption)

@DavidCarlisle Oh no!

just while output is active with a blinking cursor after that

@EmilioPisanty delete all generated file (.aux .bbl etc and start fron a clean sheet, if that doesn't work add \tracingall as late as possible (just after \bibliography perhaps) and then see what it does

@AlanMunn We are very hard to please regarding food :)

@egreg But pineapple is the worst. :)

@UlrikeFischer Oh no!

@DavidCarlisle what do you mean by clean sheet?

just the tex file on its own in a new directory?

@EmilioPisanty sorry, I just meant delete the aux and other files.

@AlanMunn I should have written “Wurst”.

@PauloCereda which airport is more convenient in SP?

@egreg You haz mail. (I need to buy all my tickets ASAP as things are complicated with transfers)

with a cursor blinking at the bottom

I'm finding this hard to reconcile with your previous comment

does this mean it's looping?

or does it count as a prompt?

@EmilioPisanty looping (the blinking cursor isn't tex that's a feature of your terminal window)

@DavidCarlisle ok, that's good to know

(I should obviously say that I've tried to distill things down to a MWE but have found the behaviour remarkably slippery, i.e. it's there one minute and gone the next at the slightest change, just to come back on the next edit to the caption.)

@CarLaTeX @CarLaTeX This helps explain why. ^^^

@EmilioPisanty \tracingall will reveal all:-)

@LoopSpace I think you should do it like this:

\documentclass{article}

\usepackage{siunitx}
\usepackage{fontspec}
\usepackage{unicode-math}
\setmainfont{TeX Gyre Pagella}
\setmathfont{TeX Gyre Pagella Math}
\sisetup{number-math-rm=\symrm}
\begin{document}
\SI{\pi}{s}
\end{document}

@UlrikeFischer or simpler, $\pi\,$s\bye

@yo' I'll examine the program ASAP.

@egreg ok thanks

@DavidCarlisle well, it's definitely made it more verbose

@EmilioPisanty and?

@DavidCarlisle I'm.... struggling on how to analyze the resulting log

there's a bunch of stuff and it definitely seemed to be looping, but it's hard to find hard features to pin onto

@EmilioPisanty i would guess there is a repeating pattern at the end, you could just put it here (or in a question)

@DavidCarlisle I don't see anything simple

@EmilioPisanty there are only so many ways you can loop by accident. redefining \par, being by far the most likely

if there's a repeating pattern it's multiple pages long in the log

@EmilioPisanty well just show the last few lines before you killed it

{entering math shift group (level 2) at line 427}
{changing \fam=0}
{into \fam=-1}
{changing \predisplaysize=0.0pt}
{into \predisplaysize=88.29999pt}
{reassigning \predisplaydirection=0}
{changing \displaywidth=0.0pt}
{into \displaywidth=243.96161pt}
{reassigning \displayindent=0.0pt}
\everydisplay->\check@mathfonts \the \everydisplay \bool_set_true:N \l__siunitx
_display_math_bool
% t=154.0 plus 1.0 g=272.15735 b=10000 p=150 c=100000#
% t=166.0 plus 1.0 g=272.15735 b=10000 p=100 c=100000#
% t=178.0 plus 1.0 g=272.15735 b=10000 p=100 c=100000#

there's a lot of entered on line

there's a lot of

@\par via @@43 b=0 p=-10000 d=15100
@\par via @@44 b=0 p=-10000 d=10100
@\par via @@45 b=0 p=-10000 d=10100
@\par via @@46 b=0 p=-10000 d=10100
@\par via @@47 b=0 p=-10000 d=10100
@\par via @@48 b=0 p=-10000 d=10100
@\par via @@49 b=0 p=-10000 d=100
@\par via @@50 b=0 p=-10000 d=5100
@\par via @@51 b=0 p=-10000 d=10100
@\par via @@52 b=0 p=-10000 d=100
@\par via @@53 b=0 p=-10000 d=5100
@\par via @@54 b=0 p=-10000 d=5100
@\par via @@55 b=0 p=-10000 d=100
@\par via @@56 b=0 p=-10000 d=10100
@\par via @@57 b=0 p=-10000 d=10100

there's occasional bursts of

\mathdisplay #1->\ifmmode \@badmath \else $$\def \@currenvir {#1}\let \dspbrk@c
ontext \z@ \let \tag \tag@in@display \let \label \label@in@display \SK@equation
true \global \let \df@label \@empty \global \let \df@tag \@empty \global \tag@f
alse \let \mathdisplay@push \mathdisplay@@push \let \mathdisplay@pop \mathdispl
ay@@pop \if@fleqn \edef \restore@hfuzz {\hfuzz \the \hfuzz \relax }\hfuzz \maxd
imen \setbox \z@ \hbox to\displaywidth \bgroup \let \split@warning \relax \rest
ore@hfuzz \everymath \@emptytoks \m@th $\displaystyle \fi \fi

there's occasional bursts of

{the letter a}
{blank space  }
{the letter t}
{blank space  }
{the letter a}
{blank space  }
{the letter p}
{blank space  }
{the letter a}
{blank space  }
{the letter r}
{blank space  }
{the letter b}
{blank space  }
{the letter a}
{blank space  }
{the letter f}
{blank space  }

@egreg ASAP = As Strange As Pineapple

there's a good deal of

\fix@penalty ->\ifnum \lastpenalty =\z@ \@@italiccorr \else \count@ \lastpenalt
y \unpenalty \@@italiccorr \penalty \count@ \fi
{\ifnum: (level 5) entered on line 426}
{true}
{\/}
{\else: \ifnum (level 5) entered on line 426}
{\fi: \ifnum (level 5) entered on line 426}
{\else: \ifdim (level 4) entered on line 426}
{\fi: \ifdim (level 4) entered on line 426}
{\fi: \iftrue (level 3) entered on line 426}
{\fi: \ifdim (level 2) entered on line 426}
{\hbox}
{entering hbox group (level 2) at line 426}

so what's on line 427 (you start a display math) or the currently open file as recroded by (foo.tex in the log with no matching )

@EmilioPisanty better use only \tracingmacros=1 for a start. Then the log is not so long.

line 427 is a \begin{equation}

nothing too crazy in it

@UlrikeFischer that just gets stuck with no output same as before

@EmilioPisanty yes it just makes the log shorter (you would not have the @\par via @@54 b=0 p=-10000 d=5100 lines for example

@EmilioPisanty No output at all is not really possible, you have macro execution above, they still must be there.

@UlrikeFischer well, if I give it \tracingmacros=1 it outputs Underfull \vbox (badness 2600) has occurred while output is active with a cursor blinking after it, and nothing else

so presumably it's looping?

ditto if I increase to \tracingmacros=20

@EmilioPisanty 1 is all it needs, but you need to look in the log not the console output unless you also add \tracingonline=1

ok

@DavidCarlisle ah

here it is

a simple repeating pattern

\iterate ->\setbox \flushend@@varbox@a \vbox {\unvbox \flushend@@tempbox@c \uns
kip \unkern \unpenalty \unskip \unkern \unpenalty \unskip \unkern \unpenalty \s
etbox \flushend@@tempbox@c \lastbox }\flushend@@lastskip@b \ht \flushend@@varbo
x@a \setbox \flushend@@varbox@a \vbox {\unvbox \flushend@@varbox@a \unskip \unk
ern \unpenalty \unskip \unkern \unpenalty \unskip \unkern \unpenalty }\advance
\flushend@@lastskip@b -\ht \flushend@@varbox@a \ifdim \flushend@@lastskip@b >\t
opskip \xdef \top@@section@skip {\the \flushend@@lastskip@b }\fi \ifdim \ht \fl

well... that is \loop :-)

it does indeed look like

@UlrikeFischer Thanks, I'll do that.

\def\loop#1\repeat{\def\iterate{#1\relax\expandafter\iterate\fi}%
  \iterate \let\iterate\relax}

I can't pick out anything that looks like a piece of my document, though

@EmilioPisanty ^^

@EmilioPisanty that's code from flushend.sty. Do you use it?

@UlrikeFischer yes

@DavidCarlisle Actually, what is the correct spacing? Sometimes I want to do \(complicated expression \si{m s^{-1}}\) and get the spacing right between the complicated expression and the unit.

let me try and kill it

@UlrikeFischer not again:-)

@DavidCarlisle ? Had we it earlier?

that does seem to fix the problem

@UlrikeFischer I have some history with that package:-)

it also doesn't seem to introduce any ugly bottoms?

I'm utterly mystified as to what about flushend would've caused this behaviour

... but if cutting it out fixes all problems then I'm happy to ditch it

@LoopSpace Are you sure you want to do it? What's the complicated expression consisting of?