I might have preferred to work with HTML too. If you're considering switching, there are a variety of simple templating languages that would help you avoid verbose, repetitive markup. And there are probably more tools for working with HTML as a source than DocBook.
DocBook having a poor toolchain for converting to PDF seems weird to me, though, since I thought PDF was one of the original target conversion formats. Could it be just because you're using DocBook 5.x and most applications still only support the 4.x series? Or, have you heard of
Publican? It's used by Fedora / Red Hat for their formal documentation, all of which ends up in both web and PDF formats.
~Andrew / cemjig
On Wednesday, June 15, 2016 at 10:04:17 AM UTC-6, Robin Powell wrote:
So back in the day when deciding what markup to use, I picked
DocBook because it could apparently do everything and was very
clear. That is, if you see <emphasis>foo!</emphasis>, you know
exactly what that means; clarity for editors.
To get to docbook, I started with a qausi-HTML dump of the original
CLL's .doc
It turns out, however, that there's no good, direct, free path from
DocBook to PDF; you have to go through XSL-FO (what?), LaTeX (badly)
or HTML (via Prince, which is what we're using, or other things;
they all costs money)
The irony is that as of now, *every single output format* goes
through HTML. In fact, all of them except PDF *are* HTML (ePub is
just the HTML output slightly re-jiggered).
So I could have saved a huge amount of work, if I'd known about how
bad the PDF chain was and if I'd known about Prince, by skipping
DocBook entirely.
(Indeed, one could argue that I should just take the auto-converted
HTML and drop the docbook, but actually it has some advantages; the
HTML is about 2x bigger (i.e. more verbose), which makes
things harder on editors, and we get chunking for free. But still.)