starting to think that #XSD and #REST don't mix because of #UPA. REST wants models to be open for evolution; UPA makes openness rather hard.
5
@mamund it's interesting to ponder the deeper source of the #UPA problem: it's not on the validation side of #XSD, it's the #PSVI creation.
1
@dret once again, the application of #OOP causes "problems" and, for many, results in marking the source content(msg) in error.
1
@mamund starting to wonder whether schemas should be regarded as client-side: they represent the processing capabilities of the client.
2
@dret More generally, validating outgoing messages against a schema is perfectly in line with Postel's law, for requests and responses alike
1
@stilkov i am all for conservative sending (honoring the rules), but liberal receiving has bounds, very tight ones in business settings.
2
@dret E.g. what kind of rigid validation needs to be applied to an incoming order? Acknowledging what you understood may be the best option.
1
@stilkov if you submit a bad delivery address, i might fix it for you, but should confirm with you before assuming it was what you intended.
3
Replying to @dret
@dret Again, I'm all for finding out what might possibly be wrong, and try to prevent it. Not with a stack trace from s11n code, though.

Apr 6, 2013 · 7:13 PM UTC

1
Replying to @stilkov
@stilkov validates against strict: act on it; validates against loose: try to be as #Postel as possible, in code; neither/nor: hopeless.
1
@dret Sounds like a good summary to me.