@DavidCarlisle That option should be irrelevant here because I'm not loading the class, only the package. It skips the entire document preamble.

@DavidCarlisle This causes it to skip the bit before \documentclass, as well as the preamble, when not compiling standalone i.e. when the package rather than the class is loaded. So, in this case, only the stuff between \begin{document} and \end{document} should be relevant. And that should just be a group, as I understand it. Apparently, it is a group with a space here for some reason.

@DavidCarlisle In this case, it shouldn't actually be doing anything because I removed the lines before \documentclass in the process of minimising the example. Often I have something like \pdfminorversion=7 or \PassOptionsToPackage{...}{xcolor} and those would cause an error when the file is \input without the ignore.

Just throwing this out there: the outcome of \textbf whose arg includes \par is somewhat crazy (admittedly after an error). Part of the text ends up appearing twice.

@marmot Sorry marmot, was offline.

@marmot interessting, your knowledge about palindromes ...

@BrunoLeFloch blame the latex team?

@cfr actually normally document is a fake environment and doesn't form a group at all, however by default standalone sets the content in an_h_ box so that it can crop the page to the natural width of the content, but a space at the start of an hbox is a space cf \fbox{x} and \fbox{ x }

I hope there are no Americans in the chat room...

@BrunoLeFloch Hi Bruno!

@DavidCarlisle Also Joseph

@yo' ?

@BrunoLeFloch I just got a document which uses morewrites. It has around 11.000 \@writefile in the aux-file (mostly from \index), and generates over 40.000 lines of logs (around 25.000 with \openout 3= xxx.mw, and between them around 15.000 empty lines). It takes ages to compile -- the first time I killed the compilation because I thought there is a loop ;-(.

@UlrikeFischer 11 writes shouldn't take that long (even if expressed to 3 decimal places:-)

@DavidCarlisle ;-). But as Bruno is french 11,000 wouldn't be good either. And grouping with spaces 11 000 is difficult -- I have no idea how to input a small non-breaking space here.

@UlrikeFischer 11 000 or 11 000 ?

@UlrikeFischer look depressingly similar to 11 000 in the font I'm using here:(

@DavidCarlisle well the optical is not so relevant here it is more the inner values that count. And there are rather good ;-) But it is still depressingly difficult to input such unicode even if one remember their numbers.

@UlrikeFischer C-x 8 ret thin sp :-)

@DavidCarlisle And then copy and paste into the browser? Well this I can do too. But it is a pain.

@UlrikeFischer I am sure @DavidCarlisle is running this chat inside emacs.

@HaraldHanche-Olsen is there any other way?

@DavidCarlisle I hear rumours.

@DavidCarlisle boo

@DavidCarlisle Winter Olympics 2018: USA men's hockey knocked out in shootout loss to Czechs

@LoopSpace, @UlrikeFischer, @TorbjørnT. Sneak preview of release notes for l3draw:

## l3color

This new *experimental* code provides basic user interfaces for
setting the color of material in documents. The input form is
based on that in the popular xcolor package, for example

    \color_select:n { red!30!blue }

Some features for supporting spot colors are also included:
this area is particularly complex and feedback on all aspects is
very welcome.

## l3draw

This new module provides a code-level approach to creating
graphical material. The approach take is *heavily* inspired

@JosephWright You understand you are opening Pandora's box, right? :)

@JosephWright You shouldn't do this. My projects will start to look like @PauloCereda's thesis if I redraw all ducks with l3draw. (I will love l3color, is there an interface to load the predefined color names from xcolor?).

@UlrikeFischer ooh thesis

Which reminds me: stop watching duck videos on YouTube and get back to the writing process!

@PauloCereda Oh yes

@UlrikeFischer Not at present: I've started small

@UlrikeFischer l3color is somewhat controversial on the team ...

@JosephWright I will stop the writing and wait for l3color, might be useful...

@UlrikeFischer What color names do you need? Like siunitx, I've started from the basis of 'official names only' (so CMYK/RGB)

@PauloCereda Just the same as xcolor, really, but build-in

@UlrikeFischer I'm hoping I can sell the idea quite quickly and shift to expl3 proper ...

@JosephWright varied. I normally look at the tables in xcolor and think "this one would look good for the water below the ducks and this one good for the desert" and then I load the needed options ;-).

@JosephWright Imho built-in color support is a very good thing. It is (to repeat myself) simply a pain to fight against xcolor-options clashes ...

@UlrikeFischer Yes I do know

@UlrikeFischer There's a reason I'm working on 'team supported' versions of existing ideas: the interfaces all end up compatible

@JosephWright another curiousity in siunitx font setup:

\documentclass{article}

\usepackage{siunitx}

%disables bold si-units:
\DeclareMathAlphabet{\mathrm}{T1}{futs}{m}{n}%

\sisetup{
    detect-all,%
    detect-inline-weight = math
}

\begin{document}


\mathversion{bold}
$1234 \num{1234} \SI{123}{\km}$

\end{document}

@JosephWright that's a stripped down version of the question with \bm. So there is a problem even without it, simply by loading fourier (which redefines \mathrm).

@UlrikeFischer From memory, for fourier I recommand math-rm = \mathnormal as it has an odd set up

@JosephWright When I add this to my example I get this:

which isn't quite perfect too ;-).

@UlrikeFischer Yes

@UlrikeFischer The problem starts with the font set up: I'll check whether my v3 code helps here (need to get that finished ...)

@JosephWright I can perhaps look at it in the evening.

@UlrikeFischer I suspect this is one I can only fix properly in v3: it really does need a completely different approach to font selection

@PauloCereda such a gentle game, football better stick to cricket

@DavidCarlisle ooh that game

@DavidCarlisle or croquet?

@HaraldHanche-Olsen field full of people wielding large hammers? sounds dangerous to me.

@DavidCarlisle and bats?

@DavidCarlisle That's nothing: a friend of mine once ejected 15 players (out of 20 total) in a basketball match, for fighting. The game was not forfeited, according to the rules: it ended playing 3 against 2.

@DavidCarlisle: Italians endorse the event, you should go. ^^

@PauloCereda it's easy for them they more or less speak the language already

@DavidCarlisle ooh

@DavidCarlisle or perhaps they want to endorse pizza

@PauloCereda -- start writing. leave notes for places you might want to go back to. don't get sidetracked. lecture complete.

@barbarabeeton ack. :)

@PauloCereda Let the feature requests roll in ...

@JosephWright :)

@PauloCereda I think I have a talk for TUG2018 ;)

@JosephWright Woo

@JosephWright: I don't see your name here: tug.org/tug2018/program.html

:D

@PauloCereda Shouldn't be too long ...

@PauloCereda Don't see yours either ;)

@UlrikeFischer Yeah, morewrites is long due for a rewrite.

@JosephWright I am hidden. :)

@BrunoLeFloch ooh Bruno is here

@PauloCereda Hi Paulo! I won't stay, sorry: I have calculations to attend to.

@BrunoLeFloch Hi Bruno! No worries, pal. :) Take care!

@JosephWright I tend to have no words at all on slides! Really? This is kind of amazing.

@AlanMunn To be fair, titles are often useful as markers to the students: I produce a handout version with notes

@JosephWright Ah, ok. That makes more sense. Are your non-wordy slides formulas and calculations or experimental results (or both?).

@AlanMunn For teaching, mainly structures and schemes: to non-experts, 'formulas'. For research, that plus figures of results

@JosephWright :) I knew that term would get me into trouble.

@AlanMunn ^^ Typical teaching slide

@JosephWright Is that all done with chemfig?

@AlanMunn Picture mode or l3chemistry (to be released in TUG 2018). :)

@AlanMunn Er, no

@AlanMunn 'A well-known commercial chemistry graphics program'

@JosephWright btw, interesting choice of caption font in the chemfig docs. :)

@JosephWright Chem... Draw?

@PauloCereda Might well be

@JosephWright: YOU ARE ON TEH LIST

@JosephWright Just tried building latex3. The directory l3rand doesn't have a build.lua in it so it failed at that point.

@LoopSpace l3draw?

@JosephWright Top one is hobby in TikZ, lower one is hobby in l3draw. I think that's worked surprisingly well.

@DavidCarlisle Yup.

@LoopSpace impressive, but of course not as impressive as this:

have other people been having issues updating TeXLive on Windows 10?

Just tried and got an error 3 times, couldn't unpack a file

When trying to update the TeXLive manager

Even running it as admin

@Canageek the mirror I use at warwick has had no updates for a few days I'm suspecting something a bit odd

@DavidCarlisle I've not updated in months

@DavidCarlisle agreed. That is impressive.

@LoopSpace you're easily impressed, it seems:-)

So, I get the standard "Warning, the TeX Live Manager [...] needs to be updated before..."

So I hit OK and press 'Update the TeX Live Manager"

@Canageek looks like tlmgr update --self -- on unix command line. I don't know how to do that on Windows however

@ChristianHupfer Likely

@DavidCarlisle Drawing a plane takes real skill.

@ChristianHupfer You just press the button and it freezes for a while then you ih tOK

;)

!!!! OK, problem fixed

It was an issue with the reposity I was using it looks like. I switched to "Standard Net Repository" and everything is fine again

@Canageek GUI - users :-P

@ChristianHupfer I compile from the command line. And I'm running apt-get update on my linux subsystem right now

OK, U Washington is in Seattle, that is only a few hours away, I'll use that mirror

I miss being RIGHT BESIDE U Waterloo's reposity, I could download all of texlive in well under 30 minutes instead of 2 hours

@Canageek No, just tried and it worked fine and updated 9 packages.

@Canageek I have a fast connection for few weeks now and instead of using it for downloading Linux distributions etc I only use it for Netflix ;-)

@UlrikeFischer Working fine for me now.

@Canageek my issue is presumably unrelated (warwick says no updates, but I just switched repository to oxford and now im getting 59 updates)

So yeah, I think my issue is probably something with the one I was using, I'll look it up once I'm done installing 246 updates

They've all downloaded, but this isn't a fast laptop

(I have no idea why it doesn't download and unpack in parallel? Download into RAM, unpack and then save to disk? It seems to be saving all to disk, then unpacking.....the disk is the slowest part of the PC, doubly so on cheap laptops like mine.....)

@Canageek I think it's been asked before, probably more robust on unknown architectures, also possibly for most people the download time rather than disk access is the limiting factor, but tlmgr is Norbert's baby

@DavidCarlisle Oh, it is like 90% of the time. And yes, Amdel's law. But still, it would remove the annoying 10 minute delay after downloading all the updates while it installs them

@DavidCarlisle (And again, if you want to tackle the biggest source of slowness, a tool like the one debian has that downloads one specific package from every repository and then sets your default to the fastest one would be realllly nice.)

@Canageek debian of course has the benefit of only needing to target debian, tlmgr perl script has to work on all kinds of machines with all kinds of operating systems so keeping it relatively simple possibly makes sense for the person maintaining it.

@DavidCarlisle I mean, is wget + a time command that hard?

@DavidCarlisle You've already got the download package code installed, and it already gives you an estimate of how long your download will take, so you've already got the time code.....

@Canageek there was a long thread recently on just getting wget to work on all texlive supported platforms, so yes:-)

@DavidCarlisle I mean, it already downloads the packages and keeps time, so those must already work, right ;)

@Canageek whatever. I just chatted with you while I installed 59 packages, if it had installed in half the time, would it have been better?:-)

@DavidCarlisle When I discover I need a new version of texlive to make a weird bug go away, when I'm on a deadline and just need to fix that one thing before I go to bed? Yes

@DavidCarlisle Or when I'm setting it up at work and we want to get on with our research?

@DavidCarlisle (It is still installing the packages on my system. On 194 of 246)

@Canageek when i was a lad, ordering a tape of the tex sources from the states took a week, compiling it took over a day, so I'm reasonably forgiving if tlmgr makes my disk spin (or whatever ssd do) for a bit.

@DavidCarlisle To be fair, I opened my ubuntu subsystem and it has been updating since before I started updating texlive

Seems to have the same disk-access bottleneck

aka my laptop sucks

I'd get a new one, but money issues (I'm getting a roomate to help with that) and well

I can't find one I like. Thinkpads haven't been decent in ages, and no one else still makes the touchpoint button

I might have to move to a trackball

@Canageek Update more regularly. Then you have only 5-10 packages.

@Canageek: I update daily... so this is about 5 packages each time

I've not been using LaTeX in a while, I've been in the lab or collaborting with non-tex users

@Canageek Blasphemy!

@ChristianHupfer Mostly just not writing, period. Having issues with my thesis work, so nothing to publish

Just the one 3D printing paper

Plus I got tenndonitus in my thumb, and I can't write LaTeX by voice command

@DavidCarlisle @JosephWright github.com/loopspace/hobby/commit/…

I did my entire first draft last paper dictating into my phone into Google Documents, then editing it on my PC and getting a co-author to format it

@Canageek What is a tenndonitus? Never read or heard this word yet...

@ChristianHupfer Sorry, bad spelling: tendonitis, tennis elbow, but in my thumb

@Canageek Ah, tendenitis.... I hope it's better now with your thumb

@ChristianHupfer Was better in early fall, or so I thought. At Christmas I got sick and tried playing video games again, and it came back. I bought Dragon Natruallyspeaking right away, and it seems better, but I'm being careful with it. Dr says tendons need to be worked to heal, so I'm trying not to baby it

It isn't NEARLY as bad as when I got RSI in high school.

That was daily pain, even when I wasn't using it, and was a year of physio just to get it to stop hurting, and then a year before it was fully healed, and I think it took some time after that to be where it was before

@LoopSpace Didn't we zap that ages ago

@LoopSpace Cool. I see a few use of internal interfaces, which perhaps might be covered by equivalents of \pgfq... (I'm unsure on whether to cover that area). I'll dig through properly later.

@JosephWright I wasn't sure how to proceed with this. Since hobby stores points and runs the algorithm after all the points are in place, it makes sense to apply any transformation as the points are declared. This means that when the algorithm actually renders the path, it should call the underlying draw commands which don't process their input. Looking at l3draw, that seemed to be the internal commands. An alternative would be to reset the transformation, but that could get messy.

@JosephWright I'm new to running l3 from source so maybe I did something wrong. I ran texlua ./build.lua install and it worked until it hit l3rand.

@LoopSpace The point functions don't apply a transform: the only processing is to expand the input from { x , y } (or a function generating that) to the internal {x}{y} form. What do you use in the pgf case?

@LoopSpace Sounds like a 'stale' folder from an old checkout (also, wasn't l3rand in l3trial anyway)

@JosephWright Yup, that's what it was. git clean -fd sorted it out.

@LoopSpace Phew :)

@JosephWright But \draw_point_transform:nn does apply the transformation. So I want that to run when the point is specified, but to not run when the path is rendered. So \hobby_lthree_addpoint:n uses \draw_point_transform:nn (incidentally, a better naming scheme must exist - suggestions?), but \hobby_lthree_moveto:nnn doesn't. So the latter needs to call the underlying moveto command without invoking the current transformation.

@LoopSpace I got that once I thought about your comment. What do you use in pgf? I can always add public interfaces with no transform (something like \draw_point_moveto_notrans:n would be a possible), but perhaps you use the pgfq functions?

@JosephWright In the pgf case, I set \pgf@x and \pgf@y directly, which is the equivalent of \pgfpoint. This is different to \pgfpointtransformed which applies the transformation. The pgf functions all only care about \pgf@x and \pgf@y but don't care exactly how they are set, so my actual code is a bit different but its effect is the same as \pgfpoint.

@JosephWright The distinction between \pgf... and \pgfq... is not important here: neither applies a transformation, it's just that the former can take more complicated expression.

@JosephWright The notrans versions would be fine for me.

@JosephWright Have to end this now ... time to sleep for me. Not a bad evening's work, I think, and all in only 54 lines of code!

@DavidCarlisle I just knew that standalone said it made document a group. The box thing makes sense, but I think standalone's documentation could be a bit clearer about this. If it described it that way, I'd have stood slightly more chance of finding the culprit. Thanks again for tracking it down.

@cfr didn't occur to me to look at the doc:-) I saw from \showoutput that it was a glue item from a word space so i just ran tracingall and looked where it added the space:-)