I received a nice comment to my earlier post on JEP-0070. Oh cool, I’m #2 on google for that search.
Anyway, as you may recall, this JEP is about using Jabber as an out-of-band authentication mechanism for web, i.e. HTTP based applications. If you don’t mind, I’ll just paste the whole comment.
John Doe (wow, someone actually is named that?) said, and I quote:
I think that Mozilla needs to implement an integrated and hidden Jabber client, rather than have Mozilla speak to your Jabber client.
The reason is that in this way the user would be able to maintain seperate Jabber identities for chatting and for logging into sites (thus avoiding SPAM on his chat IM account). Also it would be easier for a user to retain anonymity in this way, by selecting a different Jabber identity for each site he visits. I believe that this is easier and more practical if the Jabber client is implemented in Mozilla.
Now I don’t know about it being necessary to have multiple jabber accounts to acheive this anonymity, but jabber embedded within Mozilla? Genius! My opinion would be not a full jabber client, but more of a middleware thing. It just sits there handling some extra communication duties that fall outside of the predominately one-way (downlink) transfer that HTTP already handles.
What do y’all think? (I’m from Georgia, “y’all” is allowed here.)
## update
Hmm, looks like #3 on that google search shows that Gajim now has JEP-0070 support. Way to go for them, now Psi needs it. ![]()

I just updated the JEP to address comments gathered in the Last Call, so go check out version 0.9.
What PSA said, and what I’m going to say…
I just wanted to chime in that adding in a javascript layer to the idea you mentioned would make it an even better idea. No more need for polling and/or keeping an HTTP connection open to push up s. It would really make the web jive.
Odd…I got an error and PSA’s comment had the same error.
If I understand this post correctly, that’s what the JabberXM project meant to do, before it and the original Jabberzilla codebase were abandoned in favor of the more practical JIM codebase.
I wish someone would just do it already.