@JosephWright ffi is usually used to call arbitrary functions in (C or even C++ or so) libraries from scripting languages such as Python (or, in this case probably, Lua)

@mirabilos yes exactly, it was just in luajittex but they've re-arranged things so it's now accessible from normal luatex if I understand the release notes correctly. We have a long term hope of being able to use the harfbuzz opentype library from luatex, and this may be one route

@egreg have you seen any problems with the new \! adjustment in aligned? hopefully the OP will come back with a clarification but.....

@DavidCarlisle I'll give a try.

@DavidCarlisle ah cool, good luck!

@egreg thanks

@DavidCarlisle I tried this one

\documentclass{article}
\usepackage[
%  alignedleftspaceyes,
]{amsmath}

\begin{document}

\begin{align}
x&=\begin{aligned}[t] &y\\&z\end{aligned}
\end{align}
\begin{alignat}{1}
x&=\begin{aligned}[t] &y\\&z\end{aligned}
\end{alignat}
\begin{flalign}
x&=\begin{aligned}[t] &y\\&z\end{aligned}
\end{flalign}
\begin{gather}
x=\begin{aligned}[t] &y\\&z\end{aligned}
\end{gather}

\end{document}

Better yet, with a rule:

\documentclass{article}
\usepackage[
%  alignedleftspaceyes,
]{amsmath}

\begin{document}
\begin{equation}
x=\smash{\vrule height3pt depth 8cm\kern-.4pt}y
\end{equation}
\begin{align}
x&=\begin{aligned}[t] &y\\&z\end{aligned}
\end{align}
\begin{alignat}{1}
x&=\begin{aligned}[t] &y\\&z\end{aligned}
\end{alignat}
\begin{flalign}
x&=\begin{aligned}[t] &y\\&z\end{aligned}
\end{flalign}
\begin{gather}
x=\begin{aligned}[t] &y\\&z\end{aligned}
\end{gather}

\end{document}

@egreg looks like a line to me

Using tikzposter for the first time (at least, for academic purposes)! Great stuff!

@PauloCereda boo: you should use picture mode

@DavidCarlisle <3

@DavidCarlisle Also to me.

@egreg probably user error, but as it's been raised I'd be happier to see an example document, to confirm one way or the other. It was a relatively scary change to make....

@JosephWright @Werner I love how one's ideas are implemented (or almost implemented), but still, in the meta post the answer by Powers says "no, never". meta.stackexchange.com/questions/247392/…

ooh some math in luatex documentation (firefox won't show it due to https errors but IE does:-) luatex-math

Friends, a quick Q:

\begin{minipage}[t]{0.5\textwidth}
\lipsum[1]
\end{minipage}%
\begin{minipage}[t]{0.4\textwidth}
\includegraphics{foo}
\end{minipage}

I am sure I need to include something before \includegraphics, so the content appears correctly.

I added \phantom{a} and it works, but I know it's wrong and I'd love to understand what I am missing so badly.

@PauloCereda the second minipage isn't doing anything very useful, what do you want the output to do?

@DavidCarlisle I wanted it both to be at the same level. :) I will write a MWE, hold on.

@PauloCereda -- minipage is a box, and therefore it stays in vertical mode. precede the first one with \noindent or \leavevmode.

@PauloCereda it is at same level in your top image, the baseline of the minipage is its top line, the baseline of the second minipage is the baseline of the image. You could use adjustbox and give \includeimage vertical alignment options such as t an dT

\documentclass{article}
\usepackage{lipsum}
\usepackage[demo]{graphicx}

\begin{document}
\begin{minipage}[t]{0.5\textwidth}
\lipsum[1]
\end{minipage}%
\begin{minipage}[t]{0.5\textwidth}
\includegraphics{ducks}
\end{minipage}

\end{document}

@barbarabeeton no latex boxes need that:-)

@DavidCarlisle -- oh, well. i guess i'm just too reactionary.

@barbarabeeton if putting \leavevmode before a latex box has an effect that's a bug.

@barbara, @David: I am still lost. :) I wrote a MWE, maybe it's clearer now.

I am looking for something that makes my minipage's contents become like this.

I put a \phantom inside the second minipage and it worked, but it is so wrong and I have no idea what to do. :)

\documentclass{article}
\usepackage{lipsum}
\usepackage[demo]{graphicx}
\usepackage[export]{adjustbox}
\begin{document}
\noindent\begin{minipage}[t]{0.5\textwidth}
\lipsum[1]
\end{minipage}%
%\begin{minipage}[t]{0.5\textwidth}
\includegraphics[width=.5\textwidth,valign=T]{ducks}
%\end{minipage}

\end{document}

@PauloCereda \usepackage[export]{adjustbox} and \includegraphics[valign=t]{image}

@PauloCereda t or T according to taste:-)

@DavidCarlisle Definitely t

@egreg actually I'd mis-remembered which was which:-)

@egreg Oh interesting! Any hints on achieving the same thing with tikzpicture? :)

@DavidCarlisle, @egreg: Thank you! <3

@PauloCereda \begin{adjustbox}{valign=t}\begin{tikzpicture}...\end{tikzpicture}\end{adjustbo‌​x}

@egreg Thanks! <3

@PauloCereda Probably there's something better with TikZ commands.

@egreg It works for me. :)

@DavidCarlisle Now I think I understood this. thanks! :)

@PauloCereda Add \vspace:

\begin{minipage}[t]{0.5\textwidth}
\vspace{0pt}
\includegraphics{ducks}
\end{minipage}

@UlrikeFischer hm interesting!

Thanks!

@PauloCereda Thanks to Gonzalo: tex.stackexchange.com/questions/68511/…

\documentclass{article}
\usepackage{tikz}
\usepackage{lipsum}

\begin{document}

\begin{minipage}[t]{0.5\textwidth}
\lipsum[2]
\end{minipage}\qquad
\begin{tikzpicture}[baseline=(current bounding box.base),scale=.2]
    \begin{scope}
        \clip (-10,-20) rectangle (10,0);
        \draw[step=2cm,gray,very thin]
            (-12,-22) grid (10,0);
    \end{scope}
    \draw [<->] (-11,-10) -- (11,-10);
    \draw [<->](0,-21) -- (0,1);
    \clip (-10,-20) rectangle (10,0);
\end{tikzpicture}

\end{document}

@egreg oh my!

@egreg: when I finally write my thesis, it will be a beautiful document! :)

@PauloCereda When? ;-) By the way, great win yesterday

@egreg Phew, we were in need of that victory!

@PauloCereda And Corinthians next Wednesday?

@egreg Don't even remember, I get nervous. :)

@UlrikeFischer (@PauloCereda) but that's like T and aligns the top on the baseline t is like having \vspace{-\ht\strutbox}

@PauloCereda VVVVV

@PauloCereda Nice colors

@egreg DON'T YOU DARE

:)

@egreg You are everywhere. Suddenly you answer an integration problem that I had a glance at :)

@mickep Too nice to let it pass

Indeed. I was more looking at the reason the other guy's substitution did not work. But your solution clearly works, so I will leave that question for now. (Sorry for being off topic in here).

@mickep We just had some other off topic discussion, starting from chat.stackexchange.com/transcript/message/35552964#35552964

@mickep by "off topic" you mean interrupting @egreg's football conversation?

Hehe, I actually meant non-TeX, but I like the relaxed style inhere. But when you take that up, egreg, what is your team?

@mickep Another striped black and white team

Juventus?

@DavidCarlisle I know but @PauloCereda wrote "I am sure I need to include something before \includegraphics, so the content appears correctly." and so I wanted to show him what you can add to get a baseline before the graphic.

@UlrikeFischer indeed. :)

@mickep :)

@egreg Newcastle?

@JosephWright just when you thought it was safe, test failure:

*** ../build/test/tlb-extraprimitives-001.luatex.tlg	2017-02-20 15:45:28.256536900 +0000
--- ../build/test/tlb-extraprimitives-001.luatex.log	2017-02-20 15:45:33.965861800 +0000
***************
*** 242,247 ****
--- 242,248 ----
  mathdir
  mathdisplayskipmode
  matheqnogapstep
+ mathitalicsmode
  mathnolimitsmode
  mathoption
  mathrulesfam

@DavidCarlisle Fail

@DavidCarlisle Where did that come from?

@JosephWright luatex experimental but Hans added it for the |T| issue I raised on the list

@JosephWright still claims to be 1.04 but today's build not yesterday

test run just finished, that's the only diff

@JosephWright @egreg @UlrikeFischer ^^^

@DavidCarlisle The align question has now an example: tex.stackexchange.com/questions/354778/… Looks like a misunderstanding regarding what an operator is.

@UlrikeFischer ah thanks I'll look

@DavidCarlisle Looks like he took it quite serious at the end. I hope one can set it by font.

@UlrikeFischer no, not as far as I can see it's deep in the math layout C code, not in the font loading

@UlrikeFischer picture coming up...

oh ignore the third T in each case, that was trying an earlier fix each line comes from $|T|$, $\left|T\right|$, $\left|T\kern\fontcharic\textfont0"1D447 \right|

@yo' ...sorry, I missed the conversation here... could you bump to the front page to catch me up? ;)

@DavidCarlisle If I ignore the third T then I have the impression it looks better in all cases when the value is 1. I wonder where the drawback is ;-)

@UlrikeFischer yes although thinking about it the fact that the third T has double spacing means it's adding the italic correction when followed by an explicit kern which doesn't sound quite right although I must check what tex does in math mode there.

@UlrikeFischer I suspect that Hans would rather set it to 0 and have the fonts use italic correction in the way Cambria seems to, which would save luatex a few cycles, but really there needs to be a clearer opentype math spec. I know Fred Wang had lots of difficulty trying to tie down exactly what the parameters are supposed to mean while setting up firefox (and then safari) to interpret this stuff)

@DavidCarlisle: I now have a dev-system too: This is LuaTeX, Version 1.0.4 (TeX Live 2016/W32TeX). But the w32tex binaries don't have the new primitive yet.

@UlrikeFischer yes you need a build from today not yesterday.

Fedora will no longer produce Alpha releases.


== Detailed Description ==
By gating Rawhide builds from landing in the compose and gating the
publication of composes on automated test results we will ensure
Rawhide will always be at Alpha quality. This will make it more
generally useful to people as a daily driver and development platform,
and mean we no longer need to go through the process of building,
testing and shipping Alpha releases. The initial testing will be
ensuring that a package is installable and that it does not break

@yo' -- i see that eqnalign has appeared on ctan. the description alleges that it makes eqnarray behave like align. one question: does that include recognizing \qedhere, to move the tombstone up onto the last line of the display, and adjusting the following vertical space?

@DavidCarlisle You are the SVN expert: is the correct way to create a tag svn copy -r <snapshot> ^/trunk ^/tags/<name> -m "<name>"? If so, any thoughts on the names and do you think I should 'make it so' or wait for Frank's feedback ...

@JosephWright yes if we want to copy the whole of trunk, that's basically what Ive done in latex2e-public (without the tags subdir) or without the -r if you want the current head

@barbarabeeton yes, I added the relevant internal macro, it's documened in the code in the pdf

@DavidCarlisle I'm thinking of properly sorting out our release history: would make certain things easier

@JosephWright you mean like annotating changes.txt going back to last century? :-)

@DavidCarlisle As I said on the team list, one might argue for separate repos if one were starting today, but we are not and other than l3build everything is handled in one block so ...

@DavidCarlisle Something like that, yes

@yo' -- thanks. (didn't retrieve the package or doc, but want to be ready for when it shows up in something submitted for publication here.)

@DavidCarlisle L3 SVN only goes to 1997

@JosephWright yes, although older history could be resurrected if needed

@JosephWright do you want to do PL2 ? (I have a zip file, but.....)

@DavidCarlisle If you've got it ready, go for it!

@DavidCarlisle I'll take your word for it

@JosephWright OK just need to untangle myself from ffi :-) (seem to be making progress)

@DavidCarlisle Cool: I'm looking over expl3 history

@DavidCarlisle Not many snapshots before I got involved :)

I tried to check another previously compilable file, and it throws the same error.

@DiaaAbidou Looks like you've been bitten by the LuaTeX post v0.85 changes

@DiaaAbidou We'll need to see an example to understand why this hasn't all been 'mopped up' in package fixes (which have happened)

@JosephWright Would you like me to post a question with MWE?

@DiaaAbidou Please: I have a few ideas as to what might be wrong but chat is not the best place to work that out

@DiaaAbidou \pdf@shellescape from pdftexcmds package?

@DavidCarlisle @JosephWright Here is my MWE tex.stackexchange.com/q/354830/2288

@DiaaAbidou thanks (not very minimal but I'll see...) (does it have to have times new roman for example, that makes it hard to test in anything but windows)

@DiaaAbidou it is pdftexcmds though the error is texlive/2016/texmf-dist/scripts/oberdiek/pdftexcmds.lua:292: in function .....

@DiaaAbidou (@JosephWright) which means I didn't do a complete job of updating it for luatex85, but also that the luatex85 package that you've loaded didn't catch something (although that only adds compatibility for tex side not lua)

@DavidCarlisle I don't quite understand, but do you mean that luatex85 has been bugged in the recent update?

@DiaaAbidou luatex85 defines lots of aliases for old names for packages that have not been updated but it can never catch everything, syntax changes and lua interface changes are not covered. give me a minute and i'll see what's dying here

@DavidCarlisle I understand. Thanks for help.

@barbarabeeton no worries. I'll be of course glad for any feedback. I'm sure there'll be things to improve

@JosephWright ^^ :(

@DavidCarlisle What's up there?

@JosephWright haven't the foggiest

@DavidCarlisle :)

@JosephWright once I fix up pdf@shellescape to be status.shell_escape it dies on line 95 of that file of Hans which is local execute = os.execute if execute then function os.execute(...) flush() return execute(...) end end saying standard out is closed

@DavidCarlisle It seems not to die if one removes the ` io.close(file)` from the luacode.

@UlrikeFischer hmmm

more reasonabl eMWE (@DiaaAbidou @JosephWright)

\RequirePackage{luatex85,shellesc}
\documentclass{article}

\usepackage{todonotes}

\makeatletter

\def\pdf@shellescape{\directlua{tex.sprint(status.shell_escape .. " ")}}

\makeatother

\directlua{
    function sort_tdo (jobname)
    local lines = {}
%    -- read the lines in table 'lines'
    for line in io.lines(jobname) do
    table.insert(lines, line)
    end
%    -- sort
    table.sort(lines)
%    -- write all the lines
    file=io.open(jobname.."sorted", "w")
    io.output(file)
    for i, l in ipairs(lines) do io.write(l, "\string\n")

@UlrikeFischer yes I see same in the MWE above. That's... odd

@DavidCarlisle It doesn't die if one moves the \IfFileExists{\jobname.tdo}{\directlua{sort_tdo("\jobname.tdo")}}{} behind \begin{document}. It looks as if the code is closing a file handler used by pdftexcomp at begin document.

@UlrikeFischer pdftexcmds is such a mess:-)

@DavidCarlisle I do not know Lua well but the error seems reasonable. The io.output with file handle parameter replaces the standard output, it seems. Maybe it needs to save the output handle and restore before you close the file file handle? oldout = io.output(); io.output(file); ... io.output(oldout); io.close(file);?

@UlrikeFischer I wonder....

@barbarabeeton @JosephWright @DavidCarlisle practical English question, please: Do you pronounce CTAN the same way as (Ernest Thompson) Seton?

@yo' I pronounce ctan as see-tan Seton being a name could be pronounced various way but I'd guess set-un

@DavidCarlisle It doesn't die if one loads epstopdf-base before todonotes ... Which reminds that I wrote Heiko some month ago about a clash in connection with pdftexcmds. But I can't look now it is on another computer.

@UlrikeFischer better to leave a github issue if you want anything to change:-)

@DavidCarlisle If I remember correctly I thought the other package was doing something wrong and only wanted his feedback. Will look tomorrow.

Has anyone ever written their thesis in DocBook? I get the weirdest ideas, sometimes...

@yo' It's pronounced Throatwobbler Mangrove ....

@wilx github.com/pavelsmolka/master-thesis-fi/blob/master/dp.xml

@DavidCarlisle Ah. Interesting.

@yo' -- like @DavidCarlisle, i pronounce it as "see-tan", but more importantly, the two syllables have almost equal stress, whereas "seton" has a strong stress on the first syllable and almost none on the second. (also, i've always pronounced that "see-tun", almost swallowing the "t"; there was a catholic high school "seton hall" in baltimore when i was growing up, and that's how it was pronounced there.)

latex base gone to ctan

@barbarabeeton @DavidCarlisle Ok thanks!