How to handle content vs. context markup?

Hello dear fellow brains.

I again and again hit the head on the "content vs. context markup" wall.

It seems to me that the tagline of XML, "single source publishing", gives me grey hair in the head.

If I tag a document for only context (e.g. 
<order>
	<name></name>
	<orderlines>
	</orderlines>
</order>

I have quite limited layout capabilities.

On the other hand, if I use e.g. the Docbook DTD to markup a book, I will have a lot of layout oriented tags, and not so many  content related tags.

So what is the experience and best practice of "content vs. context" tagging?

B'regards

Jens Jakob