Lists Home |
Date Index |
> Dare Obasanjo wrote:
>>And I'd rather have a spec written in prose than a schema any day of the week
> Joe English wrote:
> Having all three is best (formal spec, prose spec, and sample
> instances), if they're all available. But I think I'd still
> pick the formal spec if I were forced to choose just one.
Now that I'm out of smartass mode (sorry Dare), I can speak to this.
Where this bites is not so much for widely published and reviewed
specs, thought that can happen. It's for the ad-hoc (non-pejorative)
structures that businesses come up with to get something done. Often
this is done with samples rather than schemas. XSD is relatively new
technolgy on the ground, RNG is newer still. But you get samples
usually I believe, because on the ground everyone is in a hurry to
get something out the door and formal notation is a seen as a luxury
despite the savings made in coding effort. At Propylon, we consider
schemata a priority and budget the time to create them, but to
counter that, we see a lot of pre-existing data that can only be
spec'd via samples. It's quite possible that no one person actually
knows the range and domain of the XML structure in question. Usually
we ask for maximal and minimal samples and reverse engineer the
schemas from those.
Not unrelated, what does anyone think of using XSLT to specify a
transformation between two XML structures?
Bill de hÓra