Personally I like to keep as much as possible as XML for "source",
and
then use various XSLT stylesheets to generate different versions.
I've
done HTML, PDF, RTF and others this way.
I'd LOVE to have it this way for tutorials (XML->SVG), but unfortunately the mix of text and Inkscape graphics is too hard to create in XSLT, mostly because XSLT can't get an idea of how tall or wide some element is going to be, but SVG needs absolute coordinates for positioning.
I'm all for XML as source too and would love to see XML->SVG be the process as well. And as John Pybus brought up (in another post), perhaps it can be done with XSL-FO... I'm currently looking into it.
Couple questions regarding XML as source... is there any of our documentation that is currently written in xml (no the svg ones don't count)? Also, will this be handwritten XML or will the fact that OO uses XML suffice for the time being (given that it's relatively easy to extract and manipulate)?
Also, the SVG conversion of the notes on tracing is done excluding the images that were in the original. I wanted to get your opinions on how this should be handled.
1) Do we want to deal with raster images? (as it would make sense given the nature of the doc). One significant difference though is that I would of course use the actual potrace output for the results of the traced images (vs just using the images Bob already prepared).
2) In Bob's more "complex" examples he used the Botticelli Venus, and I'm going to change that unless anyone opposes. I think that vector-loving people associate it with Illustrator.
Let me know and I'll finish this off here this morning.
-Josh