It would be more rational to output to HTML5, since there are insane amounts of HTML to X converters around (APIs and tools alike). PDF or Doc from HTML is utterly trivial at this stage.
I am very tempted to try to write my next publication in HTML. However, I seriously worry about things like footnotes, code examples, floating figures and references. CSS3 seems to have support for many of these, but I wonder how well the convert-to-PDF pipeline really works, and how flexible it really is.
It's bad enough if I have to convert my original source to some other format years down the line, but it is absolutely critical that I can at least create the initial PDF correctly.
It's bad enough if I have to convert my original source to some other format years down the line, but it is absolutely critical that I can at least create the initial PDF correctly.