On 2/17/10 10:23 PM, Joe Hildebrand wrote: > > On 2/17/10 6:27 PM, "Matthew Wild" <[email protected]> wrote: > >>> This means that a client can also intentionally uncleanly disconnect, by >>> simply closing its socket and not sending </stream:stream>. If the client >>> wants special presence while gone, it just needs to send presence (and >>> probably request ack) before hanging up. >>> >> >> +1 to the above interpretation. I don't think there's anything extra >> we need to define, just clarify. > > +1. I'd add that the presence you send before hanging up might want to have > priority -1, in case the client author hadn't considered that.
So we have three things here, in order of impact on the spec: 1. A clarification regarding the difference between a cleanly closed stream (via an explicit </stream:stream>) and what Justin called a broken stream (I'd prefer a term like "unfinished"). This is a slight clarification, not even a fix. 2. A correction to the initial value of 'h'. IMHO this is a relatively small fix, one step above an erratum. 3. A new feature that enables the client to specify how long it will be disconnected so that the server will keep the client's session alive. IMHO this is a larger feature that requires some design work. I'd prefer to clean up #1 and #2 with a quick version update to XEP-0198, then turn to #3 in a subsequent version. Peter -- Peter Saint-Andre https://stpeter.im/
smime.p7s
Description: S/MIME Cryptographic Signature
