Lists Home |
Date Index |
From: "Bill de hÓra" <email@example.com>
> Even if what you're saying made any sense, I suspect it'd be
> irrelevant. The point is the OP is saying in a roundabout way 'I
> have to hack on comments and treat them as not comments to get
> something done'. That suggests:
> * an abuse of the document
> * a badly designed document
It sounds like a well-designed document to me. MS has made it clear
which information their system uses but has kept their comment
format parseable enough that someone looking at the document could
figure out a different use for the information. It looks like a sign
of prudent XML to me.
The trouble with schemas (i.e. schemas that don't support some
notion of variation or phase, i.e. everything except perhaps Schematron
and DTDs) is that they provide no real support for
information items in the important class "this may be interesting
to someone, especially a human, but we are not promising what
format it has, which locations it can go into, and we don't
want to commit that future revisions of even the current version
of our software will keep this item, and we probably don't want
to document it either: caveat emptor"
As someone else mentioned, as a schema progresses, then it becomes
clearer which information items should have been first class (elements
and attributes). But "comment conventions" (and PIs) say
"we think this is bathwater: if you think you can find babies, be it
on your own head."
So while I agree with Aleric that it would be better for potentially
useful machine-generated metadata to be in an attribute, I disagree
with him (as he seems to say) that information must be either in
(as attributes) or completely out:--
Better blatent than latent, but better latent than never.