[
Lists Home |
Date Index |
Thread Index
]
-----Original Message-----
From: John Cowan
IMHO (and I really mean MO), the chief advantage of declarative
code is that it's not procedural, and therefore provides a
cross-check.
--------------------------
So, duplicated constraint rules written in two different languages is an
*advantage*?
To my mind, that's twice the work. The chances of one differing from the
other in subtle ways are great, and trying to keep the two in sync is a
royal pain (as I know too well).
I think one should use declarative constraints for what is easily declared.
The problem with checking limericks is that you don't have all the pertinent
data that's to be constrained in the document. Not that it can't be
inferred, but it's not spelled out as document content, and XML validation
isn't good for anything that isn't strictly content.
Now, if you had a PSIS (Post Syllabilification Info Set) and a PPIS (Post
Pronunciation Info Set) thrown in there, you'd be, well, er... set. With all
that extra junk you can validate the limerick by a schema.
Aesthetic validation would be a little harder.
|
