Hi Ben, > I've read through the JIRA issue and it talks about having two messages > being sent but one of those should just be the wave message being updated > though the wave system, and a second email should be sent though the old > school system using the update to the wave using just the unread blip(s).
I assume this is wrt the original description of "The system would basically send 2 version of the wave. A "wave" version, and an "email" version. The same way we do plain text / html currently for email. So we would add an email header including a reference to the current wave, so if the recipient is using a wave-compatible service, it would know to switch to wave. Otherwise, if it does not understand the wave header, it would just ignore it and use regular email. Something like that." My reading of this is that we send a single email, but with multiple contents (with different mime-types) in the way plain text/html email messages work a la RFC 2110 (and friends). > This will require that the wave system holds a record of the official email > of the user that signed up to the wave system... > [..] Yes it will. Bruno's system involved giving any user signed up to a wave domain an email automatically (powered by postfix o.e.), which the user could then choose to have forwarded to wherever they would prefer. The advantage of this method over handling it internally, is it allows whatever form of munging the Wave server admin prefers. It also gives us the users's 'email' programatically, which works nicely when we start having it federated, as we don't need to perform cross-server lookup's of a 'true' email. Thoughts? Ali
