Is DITA a "semantic tagging" standard?

As much as DITA has the capacity to be a system for totally semantic tagging, the reality is that what happens to those tags totally depends on what kind of output you're generating, and what processing engine you're using to do it. So the answer to the question posed in the title is, "It isn't yet. But hopefully it will be, soon." This post explains why.


Here's a small example: If you're generating HTML then, with at least some processing engines, bullet lists had better be inside paragraphs, not outside. Otherwise, you may get quadruple spacing before the start of the list.

You discover that kind of thing only by generating and examining combinations of elements, and fixing the output to do the right thing. (It's called "customization". But in cases like the one above, I think that's a euphemism for "fixing it".)

Since that is the case, you can't fully determine your upstream writing style until you have determined your downstream processing engines, and until you have all of the fixes in place (or know you can make them) so that things come out right.

Note:
Css3 is said to unify HTML and PDF processing. That could make a big difference. Things I'd like to see include state-smart settings for keeping paragraphs together and adjusting their settings--so a 2-line paragraph stays with a figure, for example, or so a bullet spaces 6 points from a preceding paragraph, but 3 points from a preceding bullet. I look forward to finding out how much of that Css3 can do.