On Sat, Nov 27, 2010 at 3:41 AM, Juha Heinanen <j...@tutpro.com> wrote: > >> Let's discus the MWI case. If a PUBLISH is received by the presence >> server, a record is created and a 200ok with 'SIP-ETag' header is sent >> back. >> If a new PUBLISH is received and the PUBLISH does not have a >> 'SIP-If-Match' header that will match the previous 'SIP-ETag' header, >> a new record will be created for the same subscription. Instead of >> creating a new record, the existing record should be updated: the >> state of the voicemail box is unique per subscriber. > >> This approach is not compliant with the rfc 3903, but it make sense to >> implement the above behavior for certain event types packages like >> mwi. > > my mwi application first checks if etag exists and depending on the > result, either uses existing one or "." in the new publish.
Hello Juha, That is the correct behavior. I have seen scenarios in which PUBLISH were sent all the time without an etag (externnotify from asterisk sending PUBLISH). It works, but it creates multiple records in the presentity table. I was thinking about extending the sever behavior for this particular kind of event packages for which a single record in the presentity table is enough to maintain the event state. Regards, Ovidiu Sas _______________________________________________ Devel mailing list Devel@lists.opensips.org http://lists.opensips.org/cgi-bin/mailman/listinfo/devel
