« first day (3587 days earlier)      last day (1331 days later) » 

12:04 AM
@AlanMunn -- With such an enthusiastic reaction, I have to ask you to consider writing up a first-person report of how you set up arara and used it to get ready for your classes this fall, and how that worked under the stress of this year's unusual situation. The submission deadline for the fall issue of TUGboat is October 1, so that sounds like just about the right time to write up such a report. Provided you have time, of course.
 
 
2 hours later…
1:45 AM
I want to create a diagram of 4 squares. The first one contains an image, the second contains just text, the third square contains a tizpicture and the third only text. We have an arrow from a square on the left to a square on the right. Any idea on how to achieve this? I also want some text on top of the arrows and below the squares
I really need to do this urgently. If you really can help me out, I would appreciate it!
I would appreciate an example. That would save me a lot of time
 
2:01 AM
I'm trying to use \draw[<->,thick] (russell.center east) -- (whitehead.center west), but getting an error that says that "center" does not exist
I also tried with middle, but same error
I need some help!!
 
2:12 AM
I needed to use mid, the problem is that it's not producing what I expect, i.e. I expect in the middle, but it's at the bottom
my nodes contain images
 
 
4 hours later…
6:05 AM
@barbarabeeton I also owe you one on xparse
 
cis
@ pgfplots

What's the correct syntax for sth. like that?
scatter/@pre marker code/.append style={
   \pgfmathsetmacro{\zzz}{1+0.2},  % works not!
/pgfplots/cube/size z=\zzz
},
Mhhh... this seems to work:
scatter/@pre marker code/.code={ \pgfmathsetmacro{\zzz}{1+0.2} },
scatter/@pre marker code/.append style={/pgfplots/cube/size z=\zzz},
 
6:34 AM
@JosephWright ........ Reminds me of something :)
 
@Skillmonlikestopanswers.xyz Yes, true
@Skillmonlikestopanswers.xyz Had other stuff on the go :(
 
@JosephWright same here :(
 
7:04 AM
@JosephWright I think you sent your hook question to the wrong list ...
 
with import <nixpkgs> {};

let

  executables = [
    "a2ping"
    ... 100 hundred other executables ...
    "yplan"
  ];

  fhs = buildFHSUserEnv {
    name = "texlive-fhs";
    targetPkgs = pkgs: with pkgs; [
      fontconfig
      freetype
      perl
      tk
      wget
    ];
    runScript = "";
    profile = ''
      export PATH="/opt/texlive/current/bin/x86_64-linux''${PATH:+:$PATH}"
    '';
  };

  makeFhsWrapper = name: writeShellScriptBin name ''
    exec "${fhs}/bin/texlive-fhs" "${name}" "$@"
@JosephWright I managed to wrap the vanilla TeX Live for Nix. ^^^
@JosephWright Although I probably only need to wrap a fraction of the executables that TeX Live ships.
 
@UlrikeFischer Oh bugger
@HenriMenke Good stuff
 
 
1 hour later…
8:19 AM
Hi all, just fyi I now moved all code version control repositories of my LaTeX package to sourceforge. See sourceforge.net/u/mscharrer/profile for a list. Example: sourceforge.net/projects/filehook. Please post feature requests and bug reports there in the future. Unfortunatly the old tickets on bitbucket were not transferable.
5
 
@MartinScharrer sourceforge? Interesting choice:-)
 
8:35 AM
@DavidCarlisle there aren't many good places for mercurial projects.
 
@UlrikeFischer probably because these days mercurial isn't a good choice for a public code repository.
 
@UlrikeFischer mercurial? Interesting choice:-)
 
@DavidCarlisle Yeah, I looked at the (limited) hosting possibilites and sourceforge seemed good, also I had also an old account there.
@DavidCarlisle Yeah, hg is basically dead, killed by git
 
@MartinScharrer lots of people have old accounts on sourceforge, actually it's not quite as bad as it was, but for many years it was just impossible to use advert farm, and most projects moved off it.
 
But I like it more as it is more streamlined. git has a lot more bells and whistles which are great for large project but not needed for smaller ones
 
8:45 AM
@MartinScharrer True but as most people have not used hg (I used it a bit in the past) it is much harder for people to interact with the repository, fork it, make pull requests etc, than if you'd used gitlab or github
 
@DavidCarlisle You mean having ads to get some money?
@DavidCarlisle Forking and pull requests were in good support on the old hoster bitbucket, but of course people would still need to now some mercurial for that
 
@MartinScharrer That's OK but for years it was reallyhard to find a "download" link to get the files you wanted as most of the things that said "download" dumped unrelated stuff on you. There is advertising and there is unsafe misleading spam, and sadly it was nearer the latter. But as I say it has got better again in recent times I think
 
Maybe I will convert them to Git sometime. Sourceforge allows for multiple repositories per project so I would not need to move the hoster or create new projects
 
@MartinScharrer but anyway it was only expressing surprise at the host, I should say thanks for maintaining the files, do it however is best for you:-)
 
9:18 AM
@MartinScharrer Agreed: I've said this to some people (around the time I moved from Mercurial to Git, and moved off BitBucket)
@MartinScharrer Like @DavidCarlisle says, you do the maintaining, you get to pick the tools :)
 
@barbarabeeton ooh
 
9:38 AM
@JosephWright I first use SVN, then had for a Python module project a team member which used HG and started to use it - after some hesitation. I liked the possibilites compared to SVN. Back than both HG and Git were not that old.
I had a look at Git a while ago and found it a little more complex than HG so I stayed with the old.
 
@MartinScharrer Similar here: siunitx started off on SVN on BerliOS
@MartinScharrer I agree Git is more complicated, or at least can be: I made the shift as I felt there was Could Be Only One (the Highland issue)
 
Now that more and more companies also start to use Git it becomes more and more imperative for me as a SW engineer to master it.
 
Dec 11 '19 at 14:47, by Ulrike Fischer
@DavidCarlisle ? my git foo sticks to simple commands and "ask Joseph" for the rest.
 
SVN is still used heavily in companies for non-public code. They like the centralistic approach for configuration and security reasons.
But I really like it when I have the whole repository myself when I use the distributed VCs
 
@MarcelKrüger I see you are continuing to finding context bugs ;-). If they are corrected could we then use variable fonts with lualatex? What would be the basic syntax?
 
9:51 AM
@MartinScharrer sure I just made commit 207993 to the svn here, we use svn quite a bit:-)
 
@UlrikeFischer If you use the latest luaotfload dev branch (I pushed the ConTeXt patch for the time being), you can already test it:
\documentclass{article}
\DeclareFontFamily{TU}{sourcecode-variable}{}
\newcommand\DeclareSourceVariable[2]{%
  \DeclareFontShape{TU}{sourcecode-variable}{#1}{n}{%
    <-> \UnicodeFontFile{SourceCodeVariable-Roman.otf}{\UnicodeFontTeXLigatures axis={weight=#2};}%
  }{}%
  \DeclareFontShape{TU}{sourcecode-variable}{#1}{it}{%
    <-> \UnicodeFontFile{SourceCodeVariable-Italic.otf}{\UnicodeFontTeXLigatures axis={weight=#2};}%
  }{}%
}
\DeclareSourceVariable{ul}{200}
\DeclareSourceVariable{el}{250}
\DeclareSourceVariable{l}{300}
Basically there is a fontfeature axis which accepts as value again a fontfeature like syntax with the axis values.
 
@MarcelKrüger ^^^success.
@MarcelKrüger the website claims that the fonts can not be installed in windows, so I didn't try but using a texmf works fine.
 
@MartinScharrer I get that
@MartinScharrer We had discussion on the team about moving: you loose the 'one true history' aspect
 
@MarcelKrüger ;-). Does axis support more values beside weight?
 
@MarcelKrüger wow, looking nice
 
@MarcelKrüger Cool
 
@UlrikeFischer The supported values depend on the font and this font only has one axis.
 
raf
10:59 AM
Thank you.
And if anyone else is familiar with any good alternative to this MathB.in, please let me know. I looking for such a math pastebin where I can write such TeX notes online and share by generating a a URL.
 
@raf you haven't really said what's wrong with mathb.in, also it seems to be open source so you could host a second instance of that, otherwise you could use github pages or anything else that lets you have a url to an html document
 
quack
 
@PauloCereda lunch
 
@DavidCarlisle oh no
 
11:22 AM
@DavidCarlisle ^^
 
Any help with this?
0
Q: How can I create a diagram in tikz that contains both images, blocks of text and tikzpictures, which are all connected by arrows?

nbroI would like to draw a diagram that contains both an image, two blocks of text and a tikzpicture. More precisely, it should look as follows Image -> Rectangle with text -> Tikzpicture -> Rectangle with text Possibly, over each arrow, I want to add some text and below each image, rectangle or tik...

 
 
2 hours later…
1:10 PM
@JosephWright a working spotcolor with dvips: tex.stackexchange.com/a/560118/2388
 
@UlrikeFischer Ah ha!
@UlrikeFischer So we probably should sort it?
 
> The residents of Churchill, Manitoba, often leave their car doors unlocked in case anyone needs to take shelter from a polar bear.
2
 
1:25 PM
@JosephWright well it looks like a good starting point. But I need to figure out how ghostscript decides to set the color, currently I get either a stroking or a fill color (and it changes depending on the following) but not both. Do you know if there is postscript to force both?
 
@UlrikeFischer I notice it's not working with the color push stack (as it's raw PostScript)
@UlrikeFischer As far as I know, PostScript doesn't have a stroke/fill split: if you look at how pgf does fills, it's by tracking/using /pgffc manually, and that's the same method I've got for l3draw at present
@UlrikeFischer Probably that's undefined behaviour!
 
@JosephWright yes that is the other question.
 
@UlrikeFischer Don't think you can mix the two, so it would be 'separations are not managed by the stack' :( (Unless we can get dvips extended - something like color push ps:<PostScript>)
 
@JosephWright aren't named colors managed by the stack? How does it work?
 
@UlrikeFischer Named ones seem to work, yes
@UlrikeFischer I can do \special{color push XC@RedSpot}
@UlrikeFischer I adjusted my input as per my comment, to make life a bit easier :)
@UlrikeFischer What you can't do is have a tint amount: that needs the raw PostScript
 
1:37 PM
@JosephWright well assuming that people don't want to use lots of different tints, one could probably get away (if the stack matters) by defining XC@RedSpot0.1 etc for the tints as a work-around.
 
@UlrikeFischer Occurred to me too: we could do that in the backend I guess (it's the place where we know the detail)
 
How do I change the distance between the text over a node and the node itself (i.e. the arrow)?
So far, I have this
\node[minimum width=2cm,minimum height=1.5cm,inner sep=0pt, rounded corners=\blockroundedcorners, draw] (tikzcode) {my nice text over the arrow};
 
@JosephWright you were faster ;-)
 
@nbro an arrow is not a node... anyway, have you tried outer sep=5pt?
 
1:53 PM
@Rmano yes, that seems to solve the issue
sorry, that was an error
I meant \draw ...
btw, I have a tizkpicture which is too much shifted to the right
anyway to bring it all a bit to the left?
with scope?
 
@nbro xshift=-1cm?
 
inside the picture or outside?
cuz inside doesn't work
 
@nbro ?
 
@UlrikeFischer First, have to adjust \@@_model_separation_init:nnnnn so it's backend-dependent
 
I am already using many scopes inside the tikz picture
\begin{tikzpicture}[xshift=-5cm] does not work
nothing happens
 
1:56 PM
Once finished, a tikzpicture is a LaTeX box. So it is positioned as a (very big, normally) character. So the positioning in the paragraph or page depends on other things.
 
well, right now the tikzpicture goes out of the page and is cut off on the right
well, of course, I don't want this to happen
 
@nbro and it shouldn't. The picture is positioned where an "X" is positioned if you substitute it with the char X (well, the baseline is different --- there is a baseline option to shift it up or down).
 
@JosephWright It is not urgent ;-). But it is nice to see that there is a way. Btw: I think the color branch could be merged.
 
@UlrikeFischer Cool
 
@nbro so make it smaller, put it at the start of the line, or put it before the start of the line if needed, like \hspace*{-1cm}\begin{tikzpicure}...
 
1:59 PM
@UlrikeFischer Didn't I do that?
 
@nbro but it could be a million different things. Normally big tikzpictures are embedded in figure environment, so in that case things are different.
 
Isn't it possible to have just a scope and scale the tizkpicture?
the problem is that breaks some of the drawings
i.e. the other scopes inside are not apparently rescaled too
 
@barbarabeeton Unfortunately I'm definitely not the right person to promote arara. I really use it very rarely, and the syllabus code all came from @PauloCereda. I'm very grateful for the tool but I don't understand it well enough to write a useful article I'm afraid.
 
@nbro scale=0.8, transform shape or use \scalebox
First hit for "scaling a tikzpicture": tex.stackexchange.com/questions/4338/…
 
ok, thanks
btw, how can I have a line break inside the text part of a node?
I am trying to do
    \node[minimum width=2.6cm,minimum height=2.0cm,inner sep=0pt, rounded corners=\blockroundedcorners, draw] (textD) {\textbf{Evaluation}\\ asdasda};
but getting ane error
 
2:10 PM
@nbro You need to add an align=center key (or other align value.)
 
@nbro maybe you missed my answer to your question here: tex.stackexchange.com/a/560107/38080, I have that (and the previous outer sep) in that...
 
ha, right
thanks
 
Hmmmmm
 
@JosephWright I think you should be able to push a name on to the stack via the dvips special and then arrange the name to be the postscript required in a header file, perhaps...
 
@DavidCarlisle Sure, that's the 'make lots of names' plan
 
2:22 PM
@JosephWright yes I see there was more discussion later, just catching up:-)
@JosephWright cant you push 1 mysepcolor mypsfunction where mypsfunction is defined in a header to take a tint value and color name and do whatever is needed?
 
@PauloCereda ^^
 
@Rmano LOL
 
yo'
@Rmano haha, but the ladder looks modular, so it may be possible to remove the last crossbar
 
@DavidCarlisle No
 
omg
:P
I need another coffee :P
 
2:31 PM
@UlrikeFischer Oh, you mean the 2e code: ah, yes, we should merge that
 
@JosephWright hmm OK it is a long time since I really knew dvips color code:-)
 
@DavidCarlisle Ah, but you can push arbitrary 'stuff' so ...
\special{color push XC@RedSpot}
\special{color push 1}
\special{color push setcolor}
\put(0,0){\line(1,1){10}}
\put(0,10){\line(1,-1){10}}
\special{color pop}\special{color pop} \special{color pop}
@UlrikeFischer ^^^ No idea if this will work!
@DavidCarlisle, @UlrikeFischer Seems to work
 
@JosephWright How does one keep track of the numbers of pops needed?
 
@UlrikeFischer I was thinking that would be done by the code inserting the pushes: it will know that it's splitting into three
 
2:51 PM
Why the heck does "\node[below=1em of unlabelleddata] {Unlabelled data};" give me an error???
! Package PGF Math Error: Unknown operator o' or of' (in '1em of unlabelledda
now I name all my files with swear words. That's my level now
 
@JosephWright we would then need two different \reset@color commands and so probably two \set@color. Perhaps we should first ask if dvips can get an update ;-)
 
\usetikzlibrary{positioning}
you gotta be kidding
I am a very happy guy, certainly
 
@UlrikeFischer I wonder if the same trick works for the fill/stroke issue
@UlrikeFischer I was thinking in expl3 terms
\cs_new_protected:Npn \@@_backend_select_separation:nn #1#2
  {
    \@@_backend_select:n {#1}
    \@@_backend_select:n {#2}
    \@@_backend_select:n { selectcolor }
  }
 
@JosephWright what do you need to put on the stack here?
 
@JosephWright ah looks about right:-)
@JosephWright actually why can't you do it in a single push rgb takes arguments so I think you should be able to do color push mycolorspec a b c as long as you have defined setmycolorspeccolor postscript operator
 
3:11 PM
@DavidCarlisle I tried \special{color push XC@RedSpot 1 setcolor} and GhostSript wasn't happy
@UlrikeFischer Ideally, the same as for pdfTeX: everything, so there is no clash between drawing and non-drawing
 
@JosephWright \special{color push mycolorspec blue 1} makes Postscript blue 1 TeXcolormycolorspec so you just need to define TeXcolormycolorspec to take two values off the stack and do whatever is needed
 
@DavidCarlisle Ah right, that I did ot know
 
\special{color push mycolorspec blue 1}
a
\special{color pop}

b

\bye
TeXDict begin 1 0 bop blue 1 TeXcolormycolorspec 166
83 a Fa(a)p Black 166 183 a(b)1929 5539 y(1)p eop end
actually that would want to be /blue or (blue) as blue isn't a defined operator
 
@samcarter_is_at_topanswers.xyz a rapid question: do you know why if in a beamer srtyle (like rlog.rgtti.com/2019/09/21/make-your-own-beamer-style-ducks) I use \usefonttheme[onlymath]{serif} as a default then I am unable to change to a full-serif with \usefonttheme{serif} in a specific document?
 
@DavidCarlisle with the spotcolor definition it would be defined wouldn't it?
 
3:21 PM
@UlrikeFischer as above it's first on the stack so you need to stop it expanding until you get to the operator, thats what / does
 
@DavidCarlisle oh, a sort of \noexpand ;-)
 
@UlrikeFischer yes exact;y it quotes the name and puts the name on the stack directly rather than evaluate it. pdf uses / as part of its syntax to match but in postscript it is the quote operator
@JosephWright @UlrikeFischer no errors and you can have any colour you want as long as it is red.
\special{!
TeXDict begin /TeXcolormycolorspec {pop pop 1 0 0 setrgbcolor} def end
}
\special{color push mycolorspec /blue 1}
a
\special{color pop}

b

\bye
 
@DavidCarlisle wouldn't black be more fitting?
 
@AlanMunn -- Oh, well. It was worth a try. Hope your efforts pay off with a successful semester.
 
@UlrikeFischer yes but black being locally switched to black is a bit hard to spot
 
3:26 PM
@Rmano If you make a copy of the theme as, say seriff, the following will work:
\documentclass{beamer}

\usefonttheme[onlymath]{serif}
\usefonttheme{seriff}


\begin{document}

\begin{frame}
	abc x $x$
\end{frame}

\end{document}
 
@DavidCarlisle I thought that is the main point to using black ;-). If the operator should do blue 1 setcolor what would then be the syntax? {pop pop setcolor}? Or how does one refer to the stack?
 
@samcarter_is_at_topanswers.xyz Ah, ok. I will remove the thing from the theme, it's confusing for users. If they want serif math then it's easier to change it.
 
@UlrikeFischer the pop pop discards the two values, 1 and /blue here, so not discarding them and doing something with them would be an alternative plan.
 
@DavidCarlisle so simply {setcolor}?
 
@UlrikeFischer I was avoiding looking up the syntax required, is it in that question?
@UlrikeFischer ah, the OP uses RedSpotDef 1.0 setcustomcolor so if you use color push mycolor RedSpotDef 1.0 then define TeXcolormycolor as {setcustomcolor} it shoudl work.
 
3:36 PM
@Rmano If you only want to make sure that math is serif and leave everything else to your user:
\documentclass{beamer}

\makeatletter
  \def\beamer@tsfont{\rmfamily}
  \def\beamer@sfont{\rmfamily}
  \def\mathfamilydefault{\rmdefault}
  \beamer@sansmathfalse
\makeatother


\usefonttheme{serif}


\begin{document}

\begin{frame}
	abc x $x$
\end{frame}

\end{document}
 
@DavidCarlisle but doesn't that mean that the tint value is already in the definition? How many argument does setcustomcolor take?
 
@UlrikeFischer no passed from the special, but im just going on the example posted let me check what setcustomcolor does..
 
@samcarter_is_at_topanswers.xyz well, really the idea was to set math serif and text sans-serif by default (like the [onlymath]{serif} thingy did) but let the user completely override this. But I am changing my mind, really --- I like mathpazo more for the slides. And it's similar to the font my University more or less is using...
 
@UlrikeFischer yes it (by convention) takes a color and a tint see page 13 of adobe.com/content/dam/acom/en/devnet/postscript/pdfs/…
 
@Rmano If you switch to a totally different font family, make sure to also fix the operator font github.com/josephwright/beamer/issues/…
 
3:47 PM
@DavidCarlisle but then the color can be used directly with \special{color push XC@RedSpot}, and if we want to pass the tint value, we need some other definition in the spot-color.pro. I have to do some shopping now, will think about it later ;-)
 
@samcarter_is_at_topanswers.xyz nice to know! But it seems that mathpzo.sty is taking care of this... at least my \sin is switching ok.
It has:
\DeclareSymbolFont{operators} {OT1}{pplx}{m}{n}
\SetSymbolFont{operators}{bold} {OT1}{pplx}{b}{n}
 
4:04 PM
@Rmano Oh, a clever font - that's great!
Maybe eulervm can then be blamed for github.com/josephwright/beamer/issues/590 instead of beamer?
 
@UlrikeFischer yes it's same as rgb or cmyk you need to define how may parameters the color model has, so if you want a tint value it always needs two
@samcarter_is_at_topanswers.xyz or Ulrike
 
@DavidCarlisle That's always a solution :)
 
@samcarter_is_at_topanswers.xyz maybe!
@samcarter_is_at_topanswers.xyz poor Ulrike ;-)
@samcarter_is_at_topanswers.xyz hmm... have you seen the last pages of faq.ktug.org/wiki/uploads/MathFonts.pdf ?
 
@Rmano outch, that hurts :) (in particular the very last page)
 
The last one is comic sans math, will not post (they can ban me)
@samcarter_is_at_topanswers.xyz LOL, crossed message
@samcarter_is_at_topanswers.xyz Arial Narrow is almost worse... ;-)
 
4:20 PM
Does anyone know who the author of this pdf is?
 
\begin{filecontents*}{spot_color.pro}
TeXDict begin
/RedSpotCMYK [0 1 0 0] def
/RedSpotSpot (RedSpot) def
/RedSpotDef RedSpotCMYK aload pop RedSpotSpot findcmykcustomcolor def
/XC@RedSpot{RedSpotDef 1.0 setcustomcolor}bind def
end
\end{filecontents*}

\documentclass{article}
\usepackage{eepic}
\special{!
TeXDict begin /TeXcolorsetcolor {setcolor} def end
}
\AtBeginDvi{\special{header=spot_color.pro}}
\begin{document}
\setlength{\unitlength}{1mm}
abc
\begin{picture}(10,10)
\special{color push setcolor XC@RedSpot 1.0}
@DavidCarlisle, @UlrikeFischer ^^^
@DavidCarlisle Does that do what you meant?
 
@samcarter_is_at_topanswers.xyz I found it searching after the link in tex.stackexchange.com/a/34267/38080 failed
 
@UlrikeFischer, @DavidCarlisle If we are on the right track I can easily adjust l3backend, although I'll need a bit more time to sort l3color
 
@PauloCereda Oh cool, comes in handy as I just started learning Lisp.
 
@AlexG ooh
 
4:28 PM
@PauloCereda Yes, out of curiosity. I am collecting ancient programming languages (except Fortran). This is my second one, after PostScript.
 
@Rmano mmm, the pdf seems to be uploaded on dozens of different servers - so makes no sense to try to get the documentclass options updated...
 
@AlexG :)
@AlexG COBOL?
 
@AlexG Why not fortran?
 
@samcarter_is_at_topanswers.xyz I am forced to use it at work.
 
@AlexG Lucky you! I always had a lot of fun with it
 
4:36 PM
@samcarter_is_at_topanswers.xyz Really, I don't like it. We are working on a code mostly written in F77, no indentation, tons of Gotos and the original developpers retired quite a while ago.
 
@AlexG Next you can move on to RPG II :)
 
@JosephWright Yeah, another candidate :)
 
@JosephWright it looks ok (after I spotted that you have removed the dependency from xcolor.pro ;-). The only draw back I spotted is that every color use creates an (identical) separation object, but I don't see how we can avoid it.
 
@AlanMunn What the heck is this?
 
@JosephWright why do you add the header in \AtBeginDvi but the definition directly in the preamble?
 
4:42 PM
@AlexG Contrary to its name, which evokes "Role Playing Game", it stands for Report Program Generator en.wikipedia.org/wiki/IBM_RPG_II
 
@UlrikeFischer Just as I was adding stuff a bit at a time
 
@AlexG Still in use I'm sure in corporate legacy code around the world. I've never had the "pleasure" of using it. My brief stint in corporate coding was using PL/1.
 
@UlrikeFischer That's Not Our Fault though, surely: should be handled by GhostScript
@UlrikeFischer I'll tidy up in an expl3 version shortly (about to go shopping)
 
4:58 PM
@JosephWright sorry I was working:-) I wasn't expecting setcolor there I think, I'll look in a bit
 
@DavidCarlisle one could probably call it e.g. setspotcolor, then it is less confusing.
 
@AlexG -- you might consider exhuming PL/1.
 
@UlrikeFischer Well yes, this was a sketch
 
@JosephWright -- PL/1 is the only programming language I ever learned in an academic environment -- for linguistics. (And taught be Henry Kucera, of whom I believe you've heard.)
 
Think you meant @AlanMunn
 
5:10 PM
@JosephWright I know ;-). One will have to think about the naming anyway. But imho it looks good. If l3color supports this correctly it will be quite a feature.
 
@JosephWright -- Oops, you're correct. (There are things that are still unpredictable on my new laptop; the positioning of the cursor is entirely more sensitive than I'm used to.)
 
@UlrikeFischer I'm going to look at this and the related question of (finally) addressing fill colour in dvips using the stack
 
wow, I just learned that I'm putting code in an Arctic Code Vault.
 
@UlrikeFischer COOL!
@UlrikeFischer The more tricky part of supporting separations is the initialisation code: I'll have to work out what to do about CIELAB (I'm thinking simply 'no CIELAB using dvips at the moment)
 
@JosephWright Imho providing support for simple spot colors is quite enough -- and if some one wants more they should give us the header code ;-).
 
6:00 PM
@UlrikeFischer ooh frozen blub
 
6:12 PM
@JosephWright you end up with both setcustomcolor and setcolor and I thought you only needed one or the other (the example setcustomcolor defined at adobe.com/content/dam/acom/en/devnet/postscript/pdfs/… are defined in terms of setcolor already but I can't see where the one here is defined)
 
@JosephWright I think @DavidCarlisle is right, one can simplify the code:
\begin{filecontents*}[overwrite]{spot_color.pro}
TeXDict begin
/RedSpotCMYK [0 1 0 0] def
/RedSpotSpot (RedSpot) def
/RedSpotDef RedSpotCMYK aload pop RedSpotSpot findcmykcustomcolor def
end
\end{filecontents*}
\PassOptionsToPackage{dvipsnames}{xcolor}
\documentclass{article}
\usepackage{eepic}
\special{!
TeXDict begin /TeXcolorsetspotcolor {setcustomcolor} def end
}
\AtBeginDvi{\special{header=spot_color.pro}}
\begin{document}
\setlength{\unitlength}{1mm}
abc
\begin{picture}(10,10)
\special{color push setspotcolor RedSpotDef 1.0}
 
6:49 PM
@UlrikeFischer or combining the header and ! specials:
\begin{filecontents*}[overwrite]{spot_color.pro}
TeXDict begin
/RedSpotDef [0 1 0 0]  aload pop (RedSpot) findcmykcustomcolor def
/TeXcolorsetspotcolor {setcustomcolor} def
end
\end{filecontents*}
\PassOptionsToPackage{dvipsnames}{xcolor}
\documentclass{article}
\usepackage{eepic}
\AtBeginDvi{\special{header=spot_color.pro}}
\begin{document}
\setlength{\unitlength}{1mm}
abc
\begin{picture}(10,10)
\special{color push setspotcolor RedSpotDef 1.0}
\put(0,0){\line(1,1){10}}
\put(0,10){\line(1,-1){10}}
 
Is it possible to put an image in math mode?
More precisely, I would like to create the following $\{ x_i, y_i \}$, where $x_i$ and $y_i$ are images
So, actually, this can be an image
i.e. I could create an image like $\{ x_i, y_i \}$ where x_i and y_i are images, so this does not have to be done in math mode, actually
 
@nbro yes what did you try, \includegraphics can be used in text or math
 
I haven't tried anything
I was just wondering
Essentially, I'd like to show a dataset, but rather than using x_i and y_i, small images of the inputs and labels
Because this will be part of a diagram
I guess something like tex.stackexchange.com/a/11072/63097
 
@DavidCarlisle yes, the /TeXcolorsetspotcolor {setcustomcolor} def could go in a header file, but I guess the /RedSpotDef [0 1 0 0] aload pop (RedSpot) findcmykcustomcolor def should be done one the fly if a color is defined.
 
@UlrikeFischer yes they could be in ! and not use a file at all, my main point was getting rid of the two definitions for [0 1 0 0] and (RedSpot) which didn't seem to be adding much as they were only used in the one place
 
7:05 PM
@DavidCarlisle yes it look neater so. Do more findcmykcustomcolor commands exist? E.g. for rgb?
 
@UlrikeFischer yes I think so, as far as I can see they are officially not part of postscript that adobe doc gives example definitions, but ghostscript implements them by default (aparently because adobe illstrator assumes they are defined...) gs says the rgb one is defined:
GS>/findrgbcustomcolor where ==
true
 
Any idea on how I could increase the size of the image here: tex.stackexchange.com/a/11072/63097?
It doesn't apparently allow me to change width=#1em
never mind, [width=5em] actually worked
not sure what that # is used for
 
@nbro change {\xlogo1} to {\xlogo{10}}
@nbro #1 is the first argument to the command, normal \newcommand use.
@nbro if you change #1 to 5em the image will be the same size in display, text, script and scriptscript size so you can simplify it and get rid of the \mathchoice
 
7:23 PM
@DavidCarlisle Why x before xlogo?
well, I guess it's because of "\def\xlogo#1..."
 
@nbro it's just a name, internal helper macro for \logo it could have been \@logo or \zzzlogo but using @ would need \makeatletter so he just used x
 
> Nintendo is now the richest Japanese company in 2020
@samcarter_is_at_topanswers.xyz ^^ hmmmmm
 
I will need to use the image inside $$\{ ... image \}$$
so I would like to increase the image and the parentheses at the same time
What's your suggestion to achieve this?
 
@nbro same as usual, if you want \{ to stretch use \left\{...\right\}
 
thanks, that seems to work
btw, do you know why the images are from the middle to the top rather than centered?
I mean, if you use this code tex.stackexchange.com/a/11072/63097
 
7:31 PM
@nbro I was just coming to that you want to center th eimage, put it in a one-cell tabular or use adjustbox package which gives \includegraphics a vertical align key
 
and rather than $\logo_{\logo_{\logo_\logo}}$, you use ${ \logo \}$
 
@nbro the $\logo_{\logo_{\logo_\logo}}$ was just to show it gets smaller in subscripts (which you disabled by replacing #1 by a fixed length)
 
ha, the \adjustbox{valign=m}{} did the trick of centering them
 
@MartinScharrer ^^^^^^ :-)
 
btw, why twice "{\xlogo1}% {\xlogo1}%" in that code?
 
7:33 PM
@nbro 4 times for displaystyle, textstyle. scriptsyle, scriptscriptsyle
 
I am trying to define another input, but, for some reason, I get an error
```
\documentclass{article}
\usepackage{graphicx}
\usepackage{adjustbox}

\def\input1#1{\adjustbox{valign=m}{\includegraphics[width=#1em]{images/input1.pdf}}}

\def\inputa{{% mathord
\mathchoice
{\input1{3}}%
{\input1{3}}%
{\input1\defaultscriptratio}%
{\input1\defaultscriptscriptratio}}}

\def\input2#1{\adjustbox{valign=m}{\includegraphics[width=#1em]{images/input2.pdf}}}

\def\inputb{{% mathord
\mathchoice
{\input2{2}}%
{\input2{2}}%
{\input2\defaultscriptratio}%
{\input2\defaultscriptscriptratio}}}
Ideas?
Wait, missing the following
\begin{document}
$$\left\{ \inputa, \inputb \right\}$$

\end{document}
 
you have redefined \input and broken latex completely.
 
oh, that's weird, cuz with only \inputa it works
 
@nbro no it breaks latex.
 
but if I also defined \inputb it doesn't
 
7:37 PM
@nbro \inputb isn't a standard latex command, no harm in defining a command of that name, but defining \input is... brave
 
but I defined \input1 and \inputa, not \input
 
@nbro no you defined \input you can not have numbers in command names. \input1 is same thing as \input 1
 
ah, didn't know that
thanks
 
@nbro don't use \def, but only \newcommand. Then you can do much less harm to your document.
 
@nbro So if you want macros in a sequence, you need to use i,ii,iii or a,b,c etc. There is a way to create macros with numbers but it's not useful inside a user level document.
 
7:41 PM
@UlrikeFischer good idea: \renewcommand\input{blub, Ulrike told me this was safe}
 
@DavidCarlisle lol
 
@DavidCarlisle I wrote \newcommand!
 
@DavidCarlisle But to be fair, she did say use \newcommand!'
 
@DavidCarlisle $\hbox{\verb|\renewcommand|}\neq\hbox{\verb|\newcommand|}$
 
@AlanMunn I blame Ulrike
 
7:49 PM
@DavidCarlisle, @UlrikeFischer I've checked something in, but this is tricky work! We likely need to look at all the other backends and set up fallback to the device colour, plus I need I think to alter a few arguments (/DeviceCMYK, etc., are awkward)
 
@JosephWright what is so difficult? the setup or the use?
 
8:02 PM
@JosephWright @UlrikeFischer any thoughts on timing for adding raw option lists? I suspect it's too late for this time.
 
@DavidCarlisle does it break something that isn't broken already?
 
@UlrikeFischer it changes the interpretation of \documentclass[linkcolor=red]{article} to only put linkcolor not linkcolor=red in the unused option list. What changes that might have are hard to say, should only change warning messages but....
 
@DavidCarlisle I thought the code would only store the list unchanged in some other command?
 
@UlrikeFischer Keeping a track of everything
@UlrikeFischer I will likely take a few days to clean up slowly
@DavidCarlisle Even if it looks 'safe', I think too late
 
@UlrikeFischer see the my mail to team list, you can't tell people (or packages) to pick up the raw list for keyval processing as the \if@in tests for the unused option list die with low level mismatched brace errors if you do \documentclass[border={1,2,3}]{article} or has a non expandable command in the value so you need to discard the values before putting them in an edef
@JosephWright that's my feeling, but no response on team list, so I thought I'd ping here:-)
 
8:14 PM
If use " \draw[->, myarrow,outer sep=8pt] (box2.east) -- (box3) node[mynode] {{\large \textbf{Evaluation}}};", is there a way to shift the text a little bit to the right?
and also the arrow
by still having "(box2.east) -- (box3)"
 
@DavidCarlisle yes I saw it. But even if doesn't improve the handling of currently invalid options like border={1,2,3} doesn't it at least improve the handling of currently valid options as it doesn't eat spaces and expands everything? I mean packages could then at least implement border= 1 2 3 don't they?
 
@PauloCereda btw, it is on page 200 in the book in front of me :)
 
@UlrikeFischer yes it would be possible (and marginally safer) to just save the raw lists but I think it's not good it leads to rather hard to document behaviour, things that work work and things that don't don't, similarly I think you have to extend \PassOptionsToPackage to add to both the current and new raw lists otherwise if the package uses the new raw one it won't see the option, so it has to be more than one extra \def to store the original argument.
 
@DavidCarlisle hm, yes. But this sounds as if a raw list beside the normal list won't really work and one needs to remove the zap@space from the original list. I wonder how much would break ;-(.
 
8:35 PM
@UlrikeFischer that breaks too much:-) I think the version in the branch is OK and just putting the key in the unused option list is the right thing to do anyway. But there are so many packages patching this and it is used in every document so making a change a few days before even a dev release seems a bit brave.
 
@DavidCarlisle yes, but I wonder how we can get a setup to try this a bit more. I have the impression that some changes need a longer testing than the time we have between releases. Perhaps I should create a local latex-dev-dev ;-)
 
@UlrikeFischer See the checkins: I think we have to do each backend separately
@UlrikeFischer \tl_trim_spaces:n
@DavidCarlisle Has to be held off
@UlrikeFischer What's the current situation re. dvipdfmx and spot colours? I've lost track a bit
 
8:53 PM
@JosephWright I tried something equivalent to that once and there really were a surprising number of documents making use of the edef expansion \def\foo{a4paper,twoside} \documentclass[\foo]{article} and the like.
 
Anyone knows the answer to my question above? :P
 
@nbro there is a question and answer site for asking questions
 
yeah, right
I just thought this was an easy thing
I will ask on the site then
What about moving an equation to the top a little bit?
like 1mm with respect to its current position?
 
@nbro \vspace{-1mm} ?
 
@JosephWright well I think it is there, tug.org/pipermail/dvipdfmx/2020-July/000081.html, but imho there aren't binaries to test yet. I could try w32tex.org, but imho Akira wrote somewhere that he didn't updated xetex.
 
9:03 PM
yes, I just found it too
thanks!!
 
9:20 PM
@UlrikeFischer OK, so could go into the backend files with a hic sunt leones warning
 
@JosephWright let me check w32tex, I'm just downloading.
I see Karl commited "increase save_size to 200000 for tex4ht+l3regex".
 
@UlrikeFischer :)
@UlrikeFischer That's them using it for ... everything ... I guess
 
@JosephWright mostly imho for math, and the guy with the long equations is putting this at a hard test.
 
9:37 PM
@UlrikeFischer :)
 
@JosephWright the newest xetex from w32tex is doing something. With the example from the mailing list I see objects in the pdf:
6 0 obj
[/Indexed/DeviceRGB 3<000000ff000000ff000000ff>]
endobj
7 0 obj
[/CalGray<</WhitePoint[.95505 1 1.089]/Gamma 2.222>>]
endobj
 
@UlrikeFischer 'something' yes
@UlrikeFischer So we likely can extend there
@UlrikeFischer I'll keep working on dvips first: will get it working better then look at drawing colour again
 
@JosephWright I will have to reread the messages. And we can check if the color stack has improved.
 
@UlrikeFischer Cool
 
9:53 PM
@JosephWright I found the earlier message tug.org/pipermail/dvipdfmx/2020-June/000064.html. Basically one has to do it like with pdftex: define an object and then reference it in the special, but without adding something to the Colorspace, dvipdfmx will handle this (and one will have to check if this clash with pgf).
 
@UlrikeFischer That's OK then, I can re-purpose the PDF versions :)
 
@JosephWright yes, looks quite straightforward. I wonder if one can implement transparent now. Well something for tomorrow ;-). It's getting late.
 
@UlrikeFischer Yup: I'm going to fix one bug then call it a day
 
TeXwelt.de is at it again. Not only am I still waiting for an apology letter that the server has been vulnerable to Heartbleed even years after its discovery and all user secrets are possibly compromised, but now I found that the email notifications use a third-party tracking service without user consent which is a clear violation of GDPR.
 
10:09 PM
@nbro outer sep is better in the node option, not the draw (is passed to everything otherwise). You have a pos option for the node, pos=0.5 is default, try pos=0.4. I think it's in the first or second tutorial in the manual...
 
10:28 PM
@HenriMenke I see you've switched to GitHub Actions
 
@Rmano Maybe I never read a latex manual :P I just don't have the time :P Anyway, thanks, I will try that
 
10:59 PM
Aug 9 at 14:55, by David Carlisle
@nbro for which you rely on volunteer labour to support the system, which is fine, but no need to abuse it.
 
Yes, you're right, but searching in the docs can take longer than asking here and currently I don't have much time
 
11:32 PM
@JosephWright Yes, because it allows me to directly upload build artefacts to the status page.
@JosephWright Also the build is a tiny bit faster with Travis CI clocking in at ~31 min and GitHub Actions at ~24 min.
@JosephWright But mileage may vary depending on the time of the day and the phase of the moon.
@JosephWright What's also nice about GitHub Actions is that you have direct access to the GitHub API without manually generating an access token.
@JosephWright I haven't taken on the deployment for releases yet but I'm wondering whether I even want that. It's not too big of a deal to just build it locally and upload manually. Then I can also GPG sign the tarball.
 

« first day (3587 days earlier)      last day (1331 days later) »