-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On 10/8/09 8:22 PM, Brett Zamir wrote: > On 10/8/2009 3:34 AM, Kevin Smith wrote: >> On Thu, Oct 8, 2009 at 10:57 AM,<[email protected]> wrote: >> >>>> On 7-Oct-2009, at 03:02,<[email protected]> wrote: >>>> >>>>> I have been going through pubsub and noticed that it is really hard >>>>> (impossible?) to extend it without editing the original schema. I am >>>>> not XSD expert but as far as I know there is no way, for example, to >>>>> add one attribute from own namespace to the schema with >>>>> >>>> tools provided >>>> >>>>> by XSD. The problem is that Pubsub schema defines all elements >>>>> directly as named element and XSD allows extension or >>>>> >>>> restriction only >>>> >>>>> to named complex- or simple types. >>>>> >>>> I believe this is on purpose. If you need to add your >>>> own attributes and elements then you should put it in the item >>>> payload with whatever custom xmlns you need. >>>> >>> Item payload is not always the right place to extend the protocol >>> because in some cases the extension should not be visible to the >>> recipients. Specifying schema in unextendable way leads into a >>> situation where it is possible to extend the protocol but impossible >>> to document the extension in formal way. >>> >> I think (and someone else may contradict this later) that the >> intention is that you can extend your implementations with namespaced >> attributes if you want, and that the schemas are just infomative. >> > > While the schema has been described as being supplementary, when this > issue of namespaced attributes was brought up earlier on jdev, while a > number of people favored allowing them, there wasn't anything definitive > I saw out of the discussion, except for Peter pointing out that some > implementations might have a problem with namespaced attributes and one > might submit bug reports for them to overcome this problem. > > My personal thought is that while not explicitly allowing namespaced > attributes may in some way encourage agreements on standards, I believe > it deters more decentralized experimentation. I do believe that > implementations (and the specs correspondingly) should be required to > support more parsing of XML (at least bona fide namespacing and > namespacing of attributes) even if the spec doesn't recommend it.
The schemas are non-normative. The point of using XML is to enable extensibility. If people want to experiment with namespaced attributes or child elements of pubsub or any other XMPP extension, then go for it. That said, some XMPP implementations might not handle attribute extensions as well as they handle element extensions. BTW the latest version of 3920bis has clearer text about this. Peter - -- Peter Saint-Andre https://stpeter.im/ -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.8 (Darwin) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/ iEYEARECAAYFAkr2kSoACgkQNL8k5A2w/vxZ3gCdGkqdd/uVvP+UW1qjGRt0rcNE 1s8AoI3hE4uDpwf3HrkBuo4hIsd7ccmU =Awdj -----END PGP SIGNATURE-----
