« first day (3616 days earlier)      last day (1300 days later) » 

2:46 AM
@DavidCarlisle I will leave this message as highlighted, I thought you only recommended to use picture :-)
 
 
4 hours later…
6:34 AM
@PabloGonzálezL picture would be better of course
 
7:33 AM
@UlrikeFischer Final part of DeviceN done: I've sorted export to CMYK, etc. for hyperref and whatnot
 
@daleif I looked yesterday at lualatex-math (loaded by unicode-math). It redefines a mathtools command \MT_cramped_internal:Nn, but the redefinition is broken, so now you get as definition simply #1#2->{\ensuremath {#2}} which probably gives wrong results. Do you have any reports about that?
@JosephWright nice ;-) I won't be able to look today, we are flying back tomorrow, so I'm busy. But I think after the release is out we should think about how to coordinate this with color, xcolor, colorspace and pgf. (And I still have no idea what to do with these blend/mixin colors.)
 
 
1 hour later…
9:12 AM
@UlrikeFischer reports as in anyone complained? No, but then again I never use unicode-math.
 
9:27 AM
Based on @DavidCarlisle's suggestion, I am using
\newenvironment{ignorehbox}{%
	\begin{setlength}{\hfuzz}{5cm}%
}{%
	\end{setlength}%
}
works fine in text, but not so for floats
There I still get overfull hbox warnings: imgur.com/xd2BtwD
anything preventing this from working with floats in principle?
(oh, forceignorehbox is the same as above but with \maxdimen instead of 5cm)
 
9:39 AM
@ComFreek floats use \sloppy so "increase" \hfuzz from their normal value but here they decrease it, you could use this (no point in using a nested environment here)
\makeatletter% if not in a package
\newenvironment{ignorehbox}{%
	\hfuzz=5cm \let\hfuzz\dimen@
}{%
}
\makeatother
 
thanks, and where should I use it?
around the subfloats or within the subfloats?
 
@ComFreek that sets the "real" hfuzz, then masks it so any further settings to \hfuzz in that scope just set the scratch register \dimen@ and the real \hfuzz register is inaccessible and unchanged
 
err subfigures
 
@ComFreek either, if it is some existing scope you don't need the environment at all, just \hfuzz=5cm inside a subfigure would be enough. You don't need the masking then as you are setting it after the figure environment has done \sloppy
 
the overfull hbox error remains strangely
ah
 
9:43 AM
@ComFreek well that's what happens if you try untested code off the internet that was written answering a question with no test file:-)
 
underneath the surrounding float \hfuzz=20cm fixed it
i'll now try reducing the value
 
@ComFreek not having overfull boxes is of course better than not warning about them (and always possible)
 
if you have typographical ideas on how to fix this overfull hbox, please tell me :)
Added an image with surrounding paragraph: imgur.com/a/XUknoNv
 
@ComFreek you can define the horizontal width of figures to be 5cm wider than the textwidth so they stick in the margin if that is what you want (not an uncommon design actually) then you would only get overfull warnings on content that was bigger than that. That would have advantages in that for example centred captions would centre on the actual allocated width not the ignored smaller textwidth. Basically \addtolength\linewidth{5cm}\hsize=\linewidth` at the start of the figure
@Skillmonlikestopanswers.xyz up for helping to translate your own text?:-) github.com/learnlatex/learnlatex.github.io/issues/…
 
9:59 AM
@daleif do you have some examples where this command matters? I could then test, if it is broken (btw: nobody uses it for this, but lualatex-math could also be used without unicode-math).
 
@DavidCarlisle not sure if centering the float left and right (so that it goes across both margins) is better
from a design perspective
 
@ComFreek \begin{setlength} looks very odd. Why don't you use \setlength directly? You are already in a group.
 
bceause I wanted to have a reusable ignorehbox env
i love reusable, modular code (having a programming background)
but anyway, now I have 0 warnings and 0 errors!
wonderful
 
@ComFreek yes true but they are design choices you can implement and tex will warn you if they fail, just zapping hfuzz means that you get no warning even in a two column document and a float in the left column over-prints the right column, or in a twoside document the float overprints into the narrow right hand margin on even pages
 
yes, I do understand the consequences :)
 
10:19 AM
@UlrikeFischer It forms the basis of the \cramped command. Described in section 3.1.2 in the mathtools manual. The manual uses $2^{2^2}$ vs $\cramped{2^{2^2}}$ as an example.
 
@UlrikeFischer presumably it's using the luatex primitives to get cramped style?
 
@DavidCarlisle yes, that is the idea, but as it uses \use:c{luatex_crampedscriptstyle:D} instead of the primitive name, it doesn't use anything now, after the rename to \tex_crampedscriptstyle:D.
 
@UlrikeFischer ohh, there is now a build in version of this so mathtools does not need to do its own? Nice.
 
@DavidCarlisle @DavidCarlisle: you can try:
\documentclass{article}
% \usepackage{filehook-ltx} % with lualatex-dev
\usepackage{mathtools}
\usepackage{lualatex-math}
\begin{document}

 $2^{2^2}$ vs $\cramped{2^{2^2}}$
\end{document}
 
@DavidCarlisle while it is possible of course, but it is time consuming. I'm guessing publishers may draw with software like blender but I'm not totally sure.
 
10:29 AM
@daleif this is what lualatex-math is trying to do. It would be nice if you could add it to mathtools.
\documentclass{article}
\usepackage{mathtools}
\ExplSyntaxOn
\def\MT_cramped_internal:Nn#1#2{{\ensuremath{
        \use:c {cramped \cs_to_str:N #1} #2}}}

\ExplSyntaxOff

\begin{document}

 $2^{2^2}$ vs $\cramped{2^{2^2}}$
\end{document}
 
@UlrikeFischer erh, mathtools on its own seems to give a better result. (btw is \ExplSyntaxOn already in the kernel?)
 
@daleif Also with unicode-math? (you would need to overwrite the definition of lualatex-math again to test.)
 
10:45 AM
@daleif yes since the Feb 2020 release
 
@UlrikeFischer that was without, mathtools should also work without unicode-math. Is that cramped thing only defined under unicode-math?
 
@AlanMunn ooh gb4e counters with cleveref
 
@daleif what I meant: are the results without the patch also better if you use a unicode math font? If not unicode-math should perhaps add something.
 
@samcarter_is_at_topanswers.xyz OH MY
 
10:56 AM
@DavidCarlisle Is this how you do your realistic drawings of aeroplanes?
 
@DavidCarlisle I can't start skype. Can you see my account name? It claims it doesn't know the one I'm trying to use.
 
@UlrikeFischer want me to poke you?
 
@PauloCereda I can't login from my laptop so I won't probably see it, don't know why.
 
@UlrikeFischer that's odd
 
@UlrikeFischer do you have an example? (I hardly ever use anything other than pdflatex as it works fine for me)
 
11:07 AM
When clicking on PDF manuals on ctan, why does ctan always redirect to an http page?
 
@ComFreek I click on the documentation, it brings me the PDF
 
yes, but on the way there (2 redirects), there is one HTTP page
and I have configured my browser to issue fat warnings for HTTP pages
 
@ComFreek I don't get it, sorry
 
@ComFreek well, the link points to a CTAN mirror, so it's probably because the mirror still relies on HTTP...?
 
11:17 AM
"German documentation" links to http://mirrors.ctan.org/macros/latex/contrib/caption/caption-deu.pdf
An HTTP link!
 
@ComFreek you probably could write CTAN then
 
oh i need to register for a mailing list first, ctan.org/help/feedback
too much effort
 
ah better
thank you
 
And why should http be a problem here?
 
11:20 AM
short answer: because it's 2020
 
@ComFreek It's just a pdf about a latex package, not your online banking site ...
 
I won't reiterate the countless arguments you can find on the Internet already :)
 
@ComFreek don't remind me of this dreadful year, please :)
I SHOULD BE IN GERMANY
 
What's there to get?
 
@ComFreek Chocolate, for starters :)
3
 
yo'
11:26 AM
@PauloCereda no, in Czechia :)
 
@PauloCereda and ducks, don't forget the ducks!
 
@yo' ooh Czechia <3
Praha! <3
@samcarter_is_at_topanswers.xyz ooh ducks indeed
 
yo'
Speaking of ducks, got a new one into my collection (the one on the right):
user image
3
 
@yo' ooh which layout keyboard is this one?
 
yo'
@PauloCereda Czech + English (the standard thing here)
 
11:30 AM
@yo' Ah cool!
> Rabbits can smell their dead relatives in the feces of predators.
WHAT?!?!?
 
yo'
I like that the Czech keyboard has all the German and French accents, and also things like € (that's useful at work too)
 
@yo' I once won a laptop with a AZERTY keyboard setup with French... it was... challenging. :)
 
yo'
@PauloCereda AZERTY -- two options: one-finger type and keep AZERTY, or a session with a razor, and switch to QWERTY
 
@yo' :)
 
@PauloCereda I once had to use the laptop of a french colleague to chat with someone else. I had to manage the whole conversation without a single full stop because I could not find the .
(it did not occur to me that something so elementary might be hidden behind modifier keys ...)
 
11:45 AM
@DavidCarlisle sure, but probably not today (I'll leave a comment there).
 
@samcarter_is_at_topanswers.xyz oh my
 
someone please tell me a number between 1 and 3
 
@samcarter_is_at_topanswers.xyz 2
@samcarter_is_at_topanswers.xyz did I get it right? Or would 1.3548628 be fine as well?
 
@Skillmonlikestopanswers.xyz thx!
@Skillmonlikestopanswers.xyz no, integer was perfectly fine (I had trouble to decide between 3 things)
 
@samcarter_is_at_topanswers.xyz and with "between" did you mean including 1 and 3? Because I just named the only integer number between 1 and 3...
 
11:56 AM
@Skillmonlikestopanswers.xyz [1;3], but as you gave a valid answer, no need to repeat the experiment :)
 
12:13 PM
@daleif well basically \usepackage{mathtools}\usepackage{unicode-math} <old mathtools definition here> and then your 2^2^2 example. In any case, we will disable lualatex-math, so that it no longer breaks mathtools, and it is then up to mathtools to decide which definition is best for lualatex.
 
12:43 PM
@UlrikeFischer do you mind adding an issue to the mathtools tracker? Then I'll have a look at it at some point. That \use:c {cramped \cs_to_str:N #1} #2}} will only work if unicode-math is loaded and the format is new enough? right?
 
@daleif I don't have time now, but will try to think at it later. The \cramped-commands are primitives, they work also without unicode math (see e.g. the example above).
 
1:08 PM
@Skillmonlikestopanswers.xyz thanks
 
1:52 PM
@PauloCereda Stalker. :) Not worth releasing, but if someone finds it useful I might as well share the code.
 
@AlanMunn oh no
 
@PauloCereda Pretty funny.
 
@AlanMunn :)
@Rmano IKEA Valladolid: ikea.com/es/es/stores/valladolid Cómo llegar a tu tienda de muebles de Valladolid: Calle Me Falta un Tornillo, Arroyo de la Encomienda, 47195 Valladolid Wait a minute: "Calle me falta un tornillo"?!
Amusing street name. :)
@AlanMunn ^^ :)
 
2:07 PM
Yes --- didn't know, but it seems it was voted upon: https://www.europapress.es/portaltic/internet/noticia-internautas-eligen-me-falta-tornillo-nombre-calle-nuevo-ikea-20111114152601.html
Anyway, it's in the great tradition of Iain M. Banks' startship names... (missing him)
 
@Rmano ooh
 
how do you use new lines in a tikz node? the \\ causes an error in
\begin{tikzpicture}
\node[rotate=0] at (0.6,3) {
$\mathbb{E}(X\mid Y)$ is a function of only $Y$.\\
$\mathbb{E}(X\mid Y=y)$ = \sum_x\Pr(X=x\mid Y=y) = \sum_x x\frac{p(x,y)}{p(y)}$};
\end{tikzpicture}
 
2:46 PM
@Anush you need to tell the node how to align: node[align=left, rotate=0]
left, right, center are valid options
Or you use also text width=6cm and then it will auto break (I think)
 
@MarcelKrüger Do we still need the luacmd branch? Also, could you add some change log info?
 
@Rmano thanks!
 
3:15 PM
@JosephWright We don't (I removed it) and yes I will.
 
@MarcelKrüger Cool: I tend to use the ChangeLog nowadays to know which parts need releasing when :)
 
@Rmano Correct. (Setting text width makes the node something like a minipage, so it also lets you use display math, for example.)
 
@MarcelKrüger Is it worth a release for this code? For the speedup? I'm planning one shortly for the colour stuff, so it would fit in
@UlrikeFischer Thinking I should do ICC colour spaces and overprinting, so complete colour entirely (as in do everything colorspace can)
 
@JosephWright Done
@JosephWright In most practical documents the speedup is minor, but if you are doing a color release anyway I would go for it.
@JosephWright I want to prepare a PR for the sys_timer thing, but it would need some coordination between the l3kernel and l3experimental releases. Can I assume that are synchronized or do we have to deal with cross-version compatibility?
 
3:39 PM
@JosephWright yes, we will need it, but I think it needs coordination with the pdf resource stuff and metadata and so on (e.g. for the OutputIntents).
 
3:49 PM
@MarcelKrüger They always go together when required
@UlrikeFischer Indeed: first I need to read up on the code
 
cis
4:01 PM
@ pgfplotstable

Is this the correct way to create loop-like styles?

\pgfplotsinvokeforeach{0,2,4}{
\pgfplotstableset{every col no #1/.style={
column type/.add={>{\columncolor{gray}}}{} }, }
\documentclass[a4paper]{article}
\usepackage{colortbl}
\usepackage{pgfplotstable}
\pgfplotsset{compat=1.17}
\pgfplotstableread[]{
X       Y1     Y2    Y3       Date
1      10     0.1   100      2011-01-01
2      20     0.2   200      2012-02-02
3      30     0.3   300      2013-03-03
4      40     0.4   400      2014-04-04
5      50     0.5   500      2015-05-05
}\data

\pgfplotsinvokeforeach{0,2,4}{
\pgfplotstableset{every col no #1/.style={
column type/.add={>{\columncolor{gray}}}{}       }, }
 
@MarcelKrüger The only thing you might want to do is step the check on expl3 version in l3benchmark, so it gives an error if you get new-l3benchmark-old-expl3
@MarcelKrüger Key is that both ChangeLog files get updates, so I know to do both releases :)
 
cis
4:14 PM
Never before in your life you have taken so many measurements as here :()
 
4:56 PM
@UlrikeFischer Not sure I really follow what Javier is up to with ICC-based colour spaces
 
@JosephWright sorry we are out now - just driving on the Canal Grande- , I can look probably only tomorrow evening.
3
 
@UlrikeFischer Oh, don't let me disturb you!
@UlrikeFischer I'm (re-)playing Assassins Creed II, partly set there .. best I can do ;) (@PauloCereda)
@UlrikeFischer Have a lot of fun!
 
@JosephWright no problem - even in Venice you sometime have to wait for the "bus" ;-)
 
5:17 PM
@UlrikeFischer ooh is it an aquactic bus?
 
@PauloCereda Das Bus?
 
@JosephWright Das AcquaBus
ooh Wasserbus
German is so easy
 
@PauloCereda Just put all the English words together, swap the Y and Z?
 
@JosephWright and replace ss by ß
 
@PauloCereda :)
 
5:27 PM
@PauloCereda ööh
 
@AlanMunn øøh
 
@AlanMunn Be careful: down that road lies umlauts on n's ...
 
@JosephWright and death metal
 
@AlanMunn and bizarre gardening accidents, indeed
 
Those blokes are Swedish but you get the idea. :)
 
5:29 PM
@JosephWright :)
 
cis
@JosephWright 'der Bus' (origin: 'der Omnibus', not German but Latin)
 
@cis Yes, I know :)
@cis I guess the joke only works for non-native speakers :(
 
cis
@JosephWright Secret grammar rule: In German, all important things are masculine, so 'der'; even if it is actually a factual thing ('das').
 
@cis I know :)
@cis As a native English speaker, the whole grammatical gender things is ... weird
 
6:33 PM
@PauloCereda yes, a vaporetto, Venice - unlike seen in film - has a quite good public transport system, which doesn't ccnsist of gondolas.
 
@UlrikeFischer ooh
 
6:59 PM
@cis This rule is clearly wrong, it is "die Ente"
 
@samcarter_is_at_topanswers.xyz :)
@samcarter_is_at_topanswers.xyz My German teacher said a lot of the genders are Latin-derived
 
7:14 PM
@JosephWright No idea, I choose French instead of Latin :) Genders also seemed to be randomly swapped between French and German...
 
@samcarter_is_at_topanswers.xyz Oh indeed, there's no logic that I can see (but of course @AlanMunn is the expert here)
 
cis
But "Gender" is a (new) hate-word in Germany. There are small cliques who abuse the English language and conjure up fake English words (e.g. 'Handy'). It's not your fault.
 
@cis Fake English words is not really a new invention: beamer
 
7:31 PM
@JosephWright I think the rule in Portuguese is quite straightforward, with a few exceptions.
 
yo'
@JosephWright I don't think this is much the case for German really. For French maybe. But note that Latin didn't have a strict rule for gender <-> declension class, so this might be even more complicated
 
@PauloCereda The rule in English is simple: feminine things and ships are feminine, masculine things are masculine and non organic things other than ships and pipe connectors have no gender.
@cis handy is a real English word, it just doesn't mean phone:-)
@JosephWright 100 documents at latexcgi.xyz again today
 
cis
7:55 PM
I want my method

\def\ShowHelps{1}% 1 "yes" 0 "no"

to catch on in general.
 
@yo' Like I say, from my teacher
 
cis
8:10 PM
I have to say, I like the English language. I am not very good; I can't say why either. I like this way of phrasing.

Normally you only learn school English here, i.e. 'Oxford English'. In practice, this means that as soon as you stop talking with academics, you can't understand any word.

I particularly liked the series 'Cobra Kai' (continuation of 'Karate Kid' 1-3).
I think that's the real English, I had to watch it with subtitles.
 
8:24 PM
@cis \def isn't latex though.
 
cis
@DavidCarlisle How so? I can use def if it doesn't overwrite anything.
 
@cis You can but you shouldn't ;)
2
 
@cis but it is not latex. LaTeX is written in TeX and there is no compilation so all tex primitives are available but that does not mean that they should be used. Also defining something as 1 in that way has some "interesting" behaviour.
@cis what do you expect this will print? \def\ShowHelps{1} TEST: \ifnum 1=\ShowHelps 1 \else 0\fi
 
cis
@DavidCarlisle --------------> I used it here: tex.stackexchange.com/a/563974/46023
 
@cis so? The fact that you used it in a tex.sx post doesn't make it latex.
 
cis
8:39 PM
@DavidCarlisle It is my unproblamatic setting
\ifnum\ShowHelps=1
\tikzset{Help/.style={draw=#1,
nodes={scale=0.6, fill=yellow, draw=#1, text=black, font=\tiny, midway, above, align=left,} },
Help/.default=red,
}\else
\tikzset{Help/.style={opacity=0, overlay},
}\fi
 
@cis so you got lucky in that case, did you answer my question about the similar ifnum test I posted above?
@cis why not the much safer, simpler and more efficient:
\ifShowHelps
\tikzset{Help/.style={draw=#1,
nodes={scale=0.6, fill=yellow, draw=#1, text=black, font=\tiny, midway, above, align=left,} },
Help/.default=red,
}\else
\tikzset{Help/.style={opacity=0, overlay},
}\fi
with \newif\ifShowHelps
 
cis
@DavidCarlisle I only know \ifnumShowHelps=1 yes \else no\fi
@DavidCarlisle That's possible too - with a "newif", I think
 
@cis no try it as I posted in case you wanted the text 1 and 0 rather than yes and no, see what you get
@cis also putting in the help texts with opacity 0 seems weird, don't you want to not add them at all in the non-debug case?
 
cis
@DavidCarlisle I don't know. I think it shoild be something like that:
\newif\ifShowhelps\Showhelpstrue
\ifShowhelps
   Code if name=True
 \else
   Code if name=False
 \fi
But I want sth. with 1 and 0.
 
@cis you still didn't say what you expected to get from \def\ShowHelps{1}\ifnum 1=\ShowHelps 1 \else 0\fi the answer requires quite a deep knowledge of TeX's inner workings, seems a strange construct to want to use in a document.
 
cis
8:57 PM
This is always 0.
But I don't know, why I should use it this way. I use:
 \def\ShowHelps{1} Test: \ifnum\ShowHelps=1  1 \else 0\fi
 
@cis as I say you happen to use \ShowHelps in that way and you get lucky and it works as the value 1 but in other cases it doesn't and in any case if you want that definition you should not use \def but \renewcommand, but even with \renewcommand it is a fragile form to use in numeric tests.
 
9:25 PM
@egreg Hi, I have done a question on meta where there is your name :-) for the changed the color of the code. For my humble opinion it was better before. I not see the differences between the same operators.
0
Q: Changed again this afternoon the color of the code?

SebastianoThis afteroon I remember that there war another color for the code {}. Is it has been changed again, after the invite of @egreg, to change from blue color to the old colors to the mainteniers of the site?

 
9:42 PM
@Sebastiano it is a network wide change the syntax highlighter has changed from google prettify (which google no longer support) to highlight.js
@Sebastiano see the "Featured on Meta " links on the main site to see the announcement
 
10:01 PM
@DavidCarlisle I give you always thank you very much for your response. Thank you again for your precious collaboration. All always the best.
 

« first day (3616 days earlier)      last day (1300 days later) »