Revision f6a67901e79d8d35e6bf30f0766b2417740090b7

[Nathan Walp]

Sean Egan wrote:
> We should propagate this branch to head for 2.2.0. I'll help review
> it; it'd be great if Nathan and Etan did too.

As not-the-mentor, I'm just doing a quick review of this, but here's my 
comments (in no particular order):

- I dislike implementing experimental JEPs in libpurple

- forces buddy icons to PNGs, which will kill animated buddy icons

- the prpl now knows way too much about PNGs.  That functionality should 
be abstracted out to somewhere else, core or otherwise.

- turns minor get-info code duplication in buddy.c into massive code 
duplication.  Abstract that out to a function or something.

- for the gateway interaction stuff, we assume that anything in our 
roster w/o an '@' is a gateway.  Why don't we simply send a disco 
request to all such entries in the roster, so we can know for sure?

- reverts a fix in jabber_disco_finish_server_info_result_cb to not send 
presence until we receive the roster.  If we don't wait for the roster, 
we get presence we're not expecting, and no one shows up the first time 
you log in.

- ping.[ch] seems like overkill for such a small feature (I know that's 
not his code, but I made the same comment about the original patch when 
it was submitted in the tracker)

- inconsistent indenting throughout


-Nathan