Damn whiteboard markers :-( net loss: 100 EUR.

(well, there is hope, but only very little)

@yo' What kind of markers do you have at that price?

@JosephWright it's not the markers, it's the shirt and the jumper :-(

@yo' Ah, makes more sense

@JosephWright and that all just because the one I have is not able to hold the ink (we've got refill ones)

@yo' Yes, mine are refillable too (I've got top-up ink for some but not all of them)

@JosephWright the problem here is that the cushion (or how you call it) inside is getting old and worn. I promised myself this had been the last filling and I'm getting a new pen. If I had done so one filling earlier, I'd have been fine.

I'm afraid that my bad mood will impact my today's teaching in a very unfortunate way :-(

> Ursula K. Le Guin

Is line break allowed after Le?

@wilx Your call: I guess 'discouraged' but it will depend on the context (for example, in my CV I've basically got no text flexibility at all so have to break where I can)

@JosephWright OK, ~ it is then. :)

@yo' wow a 100 euro outfit, around here that would imply some once-in-a-lifetime event:-)

Worth listening

Of course, they like closing things ...

Came up in the mod chat room:

pakin.org/complaint

Scott Pakin of visual FAQ, symbol list etc., fame

U+2007 FIGURE SPACE -- width of a digit. Is there such a thing/width in LaTeX?

@wilx \leavevmode\hphantom{0}

@egreg Thanks. I will use that for PUNCTUATION SPACE as well. :)

@DavidCarlisle well, I wear fancy shirts and quality jumpers, that's all... But I've started considering teaching in a lab coat

Is there a breakable thin space or do I just use \,\hspace{0pt}?

@wilx Don't forget \leavevmode in front of \,; I suggest \leavevmode\,\linebreak[0]

@egreg OK. When is the \leavevmode necessary in front of \,? Only for the line breaking thin space or also for the narrow no-break space?

@wilx The definition of \, is faulty (but it's too late for changing it): if issued in vertical mode it adds a vertical space. Not a big deal, because it usually goes between words, but in some situations it can bite. ;-).

This is it, so far: github.com/wilx/project-uspace/blob/master/uspace.sty

@wilx Great work so far

@egreg Don't be so sure about 'too late to change it' (@DavidCarlisle?)

@JosephWright I was wondering....

@DavidCarlisle \protected\def\,? :)

@JosephWright in view of the "can latex change tex" question just answered... \let\savedkern\kern\let\kern\undefined\protected\def\hkern{\leavevmode\savedker‌​n}\protected\def\vkern{endgraf\savedkern} (feeling brave today)

@JosephWright safer as it's used less often \protected\def\thinspace{\leavevmode\kern .16667em }

@DavidCarlisle :)

@DavidCarlisle I was thinking of \ifvmode rather than \leavevmode, but the same idea I guess for the sensible use cases

@JosephWright what would you do in vmode, nothing?

@DavidCarlisle Yes :)

@DavidCarlisle Can argue either way as it doesn't really currently make sense in vmode

This is weird. When I try LuaLaTeX with \usepackage[a6paper]{geometry}, it renders the small page but Acrobat Reader and Evince render it as if the A6 page was printed on what seems to be A4 page or such.

...at least with my uspace-test.tex.

@wilx \RequirePackage{luatex85}

@DavidCarlisle Oh. This is like a 3rd time I hit something that is solved by this.

@wilx the geometry maintainer has been asked multiple times since last year to update (the update needed is trivial: \pageheight instead of \pdfpageheight in new luatex) but so far no update.

@DavidCarlisle I see.

@wilx that's the downside of latex/ctan success with thousands of packages, when you get something like a major luatex change that breaks them all, Hans can fix context in an evening but fixing latex requires a year of trying to contact people and asking for updates

@DavidCarlisle Maybe the ownership/maintainership of the package should not be so sacred in case of such package and a small patch should be provided/maintained by the community?

@DavidCarlisle Command `\,' invalid in vertical mode.

@yo' that's not consistent with other space commands which all do something (possibly wrong thing for \enspace )

\def\thinspace{\kern .16667em }
\def\negthinspace{\kern-.16667em }
\def\enspace{\kern.5em }
\def\enskip{\hskip.5em\relax}
\def\quad{\hskip1em\relax}
\def\qquad{\hskip2em\relax}

@yo' ^^ latex.ltx lines 1585+

@DavidCarlisle well, the argument (given we've had a time machine) is that all of them should be invalid in vertical mode...

Actually, making \kern both horizontal and vertical mode command was a wicked cruelty of DEK :D

@yo' I fixed that above did you notice (.....\let\kern\undefined....

@DavidCarlisle :) ... given we've had a time machine of course :-)

@yo' if Hans can remove/rename \pdfpageheight and break geometry, why can't I remove/rename \kern and break everything? It's not fair.

@DavidCarlisle call it responsibility

gotta go, started teaching 1 minute ago :-)

@wilx -- in many fonts, the digits (when uniform in width) are .5em wide, or \enspace. by the way, in the comments to your code, i think you want "three", not "tree".

@DavidCarlisle :)

@egreg -- wouldn't that allow a thin space to be left at the end of a line? it would be better to have the space at the beginning of the next line, where it should be ignored (but if \leavevmode is in front of it, i don't know if that holds).

@barbarabeeton \leavevmode is a precaution against using it in vertical mode; you're right: it should be \linebreak[0]\, (and \leavevmode shouldn't be necessary)

@wilx Does happen, but normally when the maintainer is not available, not simply not doing things

@barbarabeeton Yeah, it is supposed to be "three".

@wilx in most cases (LPPL, GPL etc it would in theory be allowed but it's a minefield, who would be allowed to edit other people's packages and increment version numbers etc) it's not really fair to ask the ctan volunteers to vet incoming updates by anything more complicated than checking the update is from the original submitter

@wilx I'll ping him again luatex 1.0 is out now.....

@wilx mail sent..

@DavidCarlisle Well, I think that CTAN is similar to FreeBSD Ports. There is a bunch of core volunteer developers and those have commit access to their Subversion repository (at least it was Subversion few years ago). They take or get assigned by other core developers tickets/issues and they commit user patches into the Ports SVN repo if specific port maintainer OKs 3rd party patch. If maintainer times out and the package in question is deemed important, it gets committed anyway.

But I guess CTAN is much smaller organization so it might not be useful scheme for it.

@DavidCarlisle Superb.

@wilx do submissions to freebsd have to assign copyright to the project? (that is common in larger projects, fsf routinely do that) which makes these things much easier, but tex world has never done that (nor is it clear that there is any legal entity to which copyright could be assigned)

And as for the versioning scheme, much like any package repository, they add their own package version to the version of the source library/application.

@wilx er yes but what does \usepackage{uspace}[2016/11/01] do if ctan has a version dated 2016/10/20 updated by some Italian bloke, and github has your version dated with the same date or ...

@DavidCarlisle No, because they only package things with existing licenses. Patches are kept in separate files which are applied each time you try to build a Ports package, if any.

@DavidCarlisle Ah, well, I guess that does not work very well then.

@wilx not really the way tex works, nothing is compiled, source files and run time files are essentially the same thing

So it appears LaTeX/CTAN is too different to copy/paste processes from other projects. :)

@wilx yeah :) well, actually, to an extent, CTAN is the originator of many things :-)

First time gmail has made a "big error": it put two completely unrelated messages from two people who do not know each other into one conversation.

@yo' Did you learn anything interesting? :)

I'm finally using the PGFplots thing that @JosephWright gave me.

Now I just need to figure out how to rotate the whole figure to be Landscape

@Canageek \begin{sidewaysfigure} ?

@DavidCarlisle Trying that

@DavidCarlisle From Rotating. Works, but [clockwise] and [counterclockwise] do not seem to do anything

@Canageek should do, although long time since I tried, or of course if you just want the plot rotated but not the caption just use a normal figure and \rotatebox{...}

@DavidCarlisle This works. I just flipped the labels around so they faced the right way up. Just for the ESI

Hey ho!

@egreg ^^^ Conclusion definitely true for any new fonts!

@JosephWright Yeah, I thought it would be better to mention it, even though I don't like the “Conclusions” section in papers or theses.

@egreg: Saint Francis of Assisi! ❤️

@PauloCereda Yes! I should go to Assisi one of these days.

@egreg he is one of the my saints of devotion.

Two places I would love to visit are Padova and Assisi.

And done my IAEA inspection

Hey @JosephWright , I'm finally using your IR code for something (ESI)

Can't figure out how to resize the labels though

@wilx sorry, I'm afraid I don't understand what you mean...

/me wonders how hard it is to make the lines thicker in pgfplots

@egreg oy stop being rude about my answers in yours:-)

@DavidCarlisle Stop writing wrong answers. ;-)

@egreg it accurately answers the question asked and included a warning . (But being safe means it may not expand as you need, you did not give your actual use case here) you'll probably get the tick for answering a different question the OP meant to ask but it will be undeserved:-)

@DavidCarlisle :P

@DavidCarlisle Tick earned.

@egreg given not earned

Václav Havel would have turned 80 tomorrow. I miss him so much here.

@PauloCereda We'll have a party soon: Steven is less than 1500 points to 100K

@egreg Steven "Stackenglne"?

@yo' Yes!

@egreg ok! :)

@egreg Yay!

ARG.

This is not helpful. I don't know where I opened it.

Can't it print the last thing before I opened it as well?

Eff,

@Canageek I don't think TeX can tell it.

Got it. \begin{figure} can't be closed by \end{sidewaysfigure}

@Canageek no (it doesn't know) just type } in your editor and see where it highlights the matching {

@egreg Seems like it shouldn't be hard. Just store that bit until closed. If error, print it.

@DavidCarlisle It wasn't a } even.

@Canageek See? It was an error in the grouping structure, no way TeX can tell you where it was started.

@Canageek well that's odd as you should get a higher level latex error for bad environment names and internally environments are \begingroup\endgroup not {} are you sure that is the cause?

@DavidCarlisle Problem went away when I fixed that.

@DavidCarlisle I could hit back a few times in Notepad++, save, adn then diff the current version if you want

@Canageek actually something interesting...

you should get

! LaTeX Error: \begin{figure} on input line 6 ended by \end{sidewaysfigure}.

and you do if no rotating package but with that package you get

! Undefined control sequence.
\endminipage ...\expandafter \@iiiparbox \@mpargs
                                                  {\unvbox \@tempboxa }
l.8 \end{sidewaysfigure}

@Canageek which means the package is doing something it shouldn't and you may be telling the truth:-)

test file:

\documentclass{article}
%\usepackage{rotating}
\begin{document}


\begin{figure}

\end{sidewaysfigure}
\end{document}

@DavidCarlisle Went back and saved right before I made that change, got error. Hit forward once, saved with new name. No longer get error.

Diff shows that only that one line changed.

@Canageek ah I see. hmm.

@Canageek in normal environments \begin and \end eseesntially act first and check correct nesting before running the underlying command but lrbox is a "fake" environment because it wants to save the box outside the group otherwise there would be no point, and (if you are still following) sidewaysfigure uses lrbox internally to grab the content which is fine but when the nesting is wrong you fall into the trick code ltbox uses to save the box outside the group and the group nesting is...

@DavidCarlisle So to do what it is trying to do, it breaks a safety measure, so if you screw up, it is hard to figure out what you did wrong.

.. totally messed up. I may have a vague idea who implemented lrbox...

    \setbox#1\hbox{%
      \begingroup\aftergroup}%
        \def\noexpand\@currenvir{\@currenvir}%

That \aftergroup} injects a } after the current group ends but it kind of assumes that the current group ends in the right place:-)

@DavidCarlisle Was it you?

Ok, so my graphs now all work. If only I know how to make the lines thicker so you could see them. Oh well, good enough for now.

@Canageek might have been:-)

@DavidCarlisle and you almost didn't have to know about that, as I'd spent five minutes hunting for the unclosed bracket before mentioning it here, then found it right off.

My girlfreind call that the Kvetch rule

@Canageek sadly I've acquired rotating since Sebastian died, I may look to see if the error behaviour here can be improved.

@DavidCarlisle :(

Ok, adding a dozen pgfplots figures with a ton of datapoints might have been a mistake.