@UlrikeFischer interesting use of \@ifnextchar in the current kvdefinekeys:-)
@UlrikeFischer do you have a simple example of the Gray issue for xcolor? the patch is adding \XC@mod@gray{}% to \XC@cnv@gray but I'm trying to generate a natural looking failure but the test cases that fail all seem rather contrived (I want to report upstream)
I am unable to set text color using the Gray color model from the xcolor package. The M(NW)E below results in an "Illegal unit of measure (pt inserted)" error:
\documentclass[a4paper, 12pt]{article}
\usepackage{xcolor}
\begin{document}
\textcolor[Gray]{10}{text}
\end{document}
Does anybod...
@UlrikeFischer because the kernel \@ifnextchar saves the two branches in \def\reserved@[ab]{....}` so you need to double up the # but due to the call chain here the number of # doubling required seems a bit variable so simplest to do (what ltx@ifnextchar does, although not a documented difference) stick it in a token register so # is safe until used
@cis From that answer: "That raises the next issue which is that Cygwin expects when crossing drives to have a prefix /cygdrive/c for drive c: and /cygdrive/d for d: However by trimming the commands to the bone I eventually dropped all such problems". Did you try to convert your paths to paths that can be understood by the shell?
@AlanMunn I think what's even worse is documentation that tries to be multiple things, unsuccessfully. For example, The TeXbook, if you ignore all the dangerous-bend/small-print sections, actually amounts to a thin book that would be a reasonable tutorial if you were a plain ("basic") TeX user circa 1980. But it's (IMO) awkward as a reference, at explaining, and as a howto -- everything is technically in there, but only if you already know where to look.
@AlanMunn Similarly we have the 1300-page pgfmanual that starts with some nice tutorials, has some explanation, and is also a reference, but put all together, tends to scare people away with its size (plus people are often looking for howtos).
I like mathematica too and I hate the python-syntax; but I think Sage will become important for the future. It is only hard to use or to run at this moment. But we could solve that.
Hehe, the new arara can do some crazy thinks, I read. So I ask myself: 1. Could arara open a cmd / shell 2. Paste some comands in it and run them 3. Close the cmd / shell
@cis that sounds a strange way of operating, why would you want the expense of a terminal window, just write the sage commands to a file and execute them, also why do you want to run sage from within TeX rather than just use sage before starting tex?
@ShreevatsaR Yes indeed. The TikZ manual is indeed imposing. But single type manuals can also be frustraing. One that comes to mind is biblatex which is purely a reference manual and could sorely use a how-to manual as well. But since everything is volunteer it's hard to change these things.
@ShreevatsaR I like the TikZ manual, but yes, it's daunting. It might work well if one had a wall length panel monitor like in those CSI type shows, where one can keep multiple pages open at once. I'm having to constantly flip back and forth.
@FaheemMitha The main frustrating thing about the TikZ manual is that the code fragments don't tell you which TikZ library needs to be loaded to use them. Adding one comment line to each would be a big help.
@AlanMunn Well, perhaps not \begin{document} and \end{document}. But I don't think it's overkill. That way, those examples could be automatically compiled to test them. Which would be a big help to users. It's done elsewhere.
@DavidCarlisle I uploaded PGF 3.1.5 to CTAN yesterday and I'm sure the bug reports will start rolling in, so if I have to fix regressions, I don't want to have to fix the CI at the same time.
@DavidCarlisle Will you complete the oberdiek split before TL2020?
@DavidCarlisle Also, are you going to split oberdiek entirely and remove it? That would actually be a good thing.
@HenriMenke it's done unless someone reports some bugs. I am done for a bit. The 40 something packages remaining are I think not so much used and unlikely to need updates so splitting doesn't help me so much and annoys people with minimal CI builds just as much. main focus now is on reducing the dependencies for hyperref 9ideally getting them down to no contrib packages at all)
While on the topic of TikZ documentation, I think a bigger problem is that there's not much on mental models / how to approach thinking about a given task — e.g. (1) get a clear idea of what you want to draw, (2) mentally express / decompose it into nodes and paths that together will result in the picture you want, (3) then look into the shortcuts that TikZ gives you for drawing these.
At least, that's how I imagine it's done; usually most answers on this site to TikZ questions just draw the figure, without explaining the author's thinking from the moment of seeing the question to deciding how to do it in TikZ.
@UlrikeFischer no But rather than just send (another) email with general request for update, am about to send one with an explicit diff, saying "can you release this, or let us do it"
But it's generally true of software that they show you tools that you can use. They often don't show you how you can use them. At least, beyond a very minimal level.
@FaheemMitha I've already been thinking about splitting the manual into two PDFs. Tutorials and reference manual, but then you lose the nice hyperlink behaviour in code block.
@HenriMenke I do read it - but I have always problems to find the page about the |- and -| syntax as I never remember which search I used last time ;-(
@HenriMenke That chapter looks nice, but is not the kind of thing I was thinking about... to pick the latest ("active") TikZ question on the site: tex.stackexchange.com/questions/521103/… — the answer just says "you could define a pic", but nothing about how to go from the question to TikZ-world
Definitely very efficient to just post "please draw this for me" though :-)
@DavidCarlisle That's a good thing: with picture I can guess roughly how much effort will be needed, while with TikZ I have no idea whether something is going to be trivial or impossible
Another way of saying what I was saying earlier: for any drawing-by-writing-code, there are two parts: (1) The part that is common regardless of whether you're using TikZ or Metapost or Asymptote (or picture too I guess) (2) The part that's specific to TikZ/Metapost/Asymptote. Think of them like design and implementation, or pseudocode and code. The former (1) is what would help the reader become self-sufficient, but answerers are eager to post (2).
@ShreevatsaR Yes, indeed. The reputation game system works really well for these type of questions on this site.
@FaheemMitha No, I'm mostly using ConTeXt with MetaFun nowadays. However, during my undergrad I drew all diagrams in my lecture notes first with PSTricks and later with TikZ.
What is the meaning of the term "format", as used in the following comment I received to a question I posted on this forum:
LaTeX3' isn't released as a format: perhaps you mean expl3?
The answer I received to the above mentioned question also mentions this term:
You can dump a format fr...
@HenriMenke Not really that helpful. LaTeX is a collection of TeX macros. I already knew that. Is MetaFun a collection of MetaPost macros? Does MetaPost have macros, or just functions?
@ShreevatsaR -- "awkward as a reference", which is one reason TeX by Topic was written. "only if you already know where to look", fortunately, Knuth was very receptive to suggestions for additions to the index; unfortunately, it's a bit too late for that now.
@ShreevatsaR -- A "combined" manual that I like very much is Kopka & Daly. It functions well as a tutorial, but also serves as a reference long after that point.
@barbarabeeton The index is great; it's just that having to look at (say) five separate pages and understand the surrounding context etc., is a bit of work. Plus because the TeXbook tries to specify the behaviour of TeX completely down to the smallest detail, it sometimes has to explain rather obscure corner-cases, in words that aren't (cannot be) too clear.
@ShreevatsaR -- I've been fortunate that I was Knuth's bug collector for many years. (That has now been turned over to Karl Berry.) So I was the beneficiary of additional explanation by both him and the knowledgeable "assistants" that he agreed to accept as bug-checkers. Not a perfect system, but the end result is, I believe, that the TeXbook is among the most reliably accurate manuals in existence. (It does help to be old enough to recognize some obscure references.)
@FaheemMitha exactly, .ecode={\edef\assignlabel{true}}, is \edef\internalcode{\edef\assignlabel{true}} so if \aasignlable is currently say false that is \def\internalcode{\edef false{true}} so you will get a syntax error on the \edef when you use the key.
I am unable to set text color using the Gray color model from the xcolor package. The M(NW)E below results in an "Illegal unit of measure (pt inserted)" error:
\documentclass[a4paper, 12pt]{article}
\usepackage{xcolor}
\begin{document}
\textcolor[Gray]{10}{text}
\end{document}
Does anybod...
What is the meaning of the term "format", as used in the following comment I received to a question I posted on this forum:
LaTeX3' isn't released as a format: perhaps you mean expl3?
The answer I received to the above mentioned question also mentions this term:
You can dump a format fr...
