URI compliance (was: Re: Annoying Collection-Name Bug)
Pierrick Brihaye wrote:
> I know that we currently accept "unicode" collection names (see the
> "città" test cases). Should we continue to do so ? Should we go towards
> URI compliance ? Should we unescape the provided URIs ? Well.. this is
> worth a discussion I think.
An important point, with strong arguments, though of different kinds, on
both sides. "Don't break existing apps or outlaw widespread practices"
against "Comply with standards". A familiar conflict. What makes it
trickier, I think, is that the Web (and indeed the heads of quite a lot of
people who Think They Know these things) is full of rather seriously
mistaken or outdated info about encoding issues in general and uri encoding
in particular, so that at least some of those whose apps get broken are
likely not just to wince and rewrite, but to howl and denounce. Which might
divert energies that would be better applied elsewhere right now. On
balance, I'd say: yes, full uri compliance in the longer term (maybe before
the next official version release). But I'd advise against any move in that
direction in an imminent snapshot release.
> balance, I'd say: yes, full uri compliance in the longer term (maybe before
> the next official version release). But I'd advise against any move in that
> direction in an imminent snapshot release.
I want to add that URI compliance would help in defining a robust
base-uri scheme and thus in writing robust XPath documents-related
However, I would be glad to hear something against URI-compliance.