Kevin Smith wrote:
On Jan 11, 2008 11:29 PM, Peter Saint-Andre <[EMAIL PROTECTED]> wrote:Peter Saint-Andre wrote: Based on the list discussion, I have updated the provisional version of XEP-0115 (Entity Capabilities).I think: <p>A client SHOULD enable a human user to disable inclusion of the 'v' attribute, which specifies a version of the software. If the 'v' attribute is not included, the receiver MUST assume that the version is intended to be private, and MUST NOT automatically send Software Version requests to the sender.</p> would be better as: <p>A client MAY enable a human user to disable inclusion of the 'v' attribute, which specifies a version of the software. If the 'v' attribute is not included, the receiver SHOULD NOT automatically send Software Version requests to the sender, although it MAY allow Software Version requests to be sent at the request of the user.</p> (the MUSTs on the version isn't really compatible with what's already out there, and we're aiming for backwards compat, I think)
That fine with me.
I'm not sure about the i18n section, it seems better to me to leave the name in.
The text in the provisional Internationalization Considerations was simply moved from the old Security Considerations (it seemed out of place there).
By "leave the name in" do you mean changing the algorithm for generating the message ("S") that's provided as input to the hashing function? That would be a change from version 1.4 of the spec.
Right now "S" is constructed as follows (with special delimiters etc. as described in the spec):
S = identities + features Including the name(s) would mean changing the algorithm to: S = identities + features + names (or somesuch)I don't see a compelling reason to change the algorithm. Indeed, adding the name means that if, say, Psi and Gajim support the same features, they would present different hashes since the name for one might be "The Psi Client" and the name for the other might be "Gajim XMPP Client" or whatever. IMHO that's not what we're trying to accomplish with caps.
Apart from these minor tweaks, I'm ok with this spec now.
Super! Peter -- Peter Saint-Andre https://stpeter.im/
smime.p7s
Description: S/MIME Cryptographic Signature
