Re: RTF for emacs

[Yuri Khan]

On Sun, May 25, 2014 at 1:22 AM, Emanuel Berg <embe8573@student.uu.se> wrote:

> I have been looking for a way to
> do simple diagrams (trees, FSMs and so on) with the
> keyboard only (i.e., generate from definitions) - I
> have tried with ImageMagick and even Dia, as it uses
> XML, so why not edit that first hand?

I do my diagrams in a self-invented semi-human-readable markup
language[1], which is then post-processed using Python scripts into
Graphviz[2] graph description language, which is then further rendered
into any image format you like (notably PNG, SVG, PDF and EPS). My
format of choice happens to be SVG, which I can conveniently view in
my web browser, print out without any loss of quality, and/or embed in
an HTML page.

[1] http://yurivkhan.github.io/textuml/ but see also
http://sf.net/plantuml/ for a Java-based alternative
[2] http://www.graphviz.org/

One could write Graphviz directly but it is too low-level and verbose.
I view it as a graph assembly language and build higher-level tools on
top of it.

> Answer: Problem with both approaches is that it
> required too much back-and-forth
> edit-and-check-and-fix-and-check-etc. so it just wasn't
> pleasant or efficient (perhaps if you had a dual
> monitor/projector and on-the-fly-update it would be).

Yes, it involves this feedback loop, and it’s mildly frustrating. And
yes, dual monitors help very much.

> When I got a cool diagram, how do you propose I embed
> it with the LaTeX PDF?

In my opinion, nohow. PDFs are for getting your article printed, and
that’s becoming more and more irrelevant. Better publish online in
HTML with illustrations in SVG. (But if you must, Graphviz can also
generate Encapsulated PostScript which LaTeX will happily include in
the PDF.)