pidgin: ae37fe6b: Added "chat-join-failed" signal, emitted...

[Mark Doliner]

On Mon, 28 Apr 2008 23:52:23 -0400, Evan Schoenberg wrote
> On Apr 28, 2008, at 11:26 PM, Mark Doliner wrote:
> 
> > This API is a little weird... when we join a chat we give the PRPL a  
> > hash
> > table containing the chat info.  For Jabber this is the room name  
> > and the
> > server.  For AIM this is the room name and the exchange.
> >
> > But your function only takes a name.  For Jabber I guess you're  
> > passing the
> > value of what would have been PurpleConversation->name?  This is  
> > weird to me
> > because it differs from the components hash table passed to join  
> > chat.  I'm
> > wondering how you envision this working for AIM?  Or, what would  
> > happen if you
> > try to join "testroom" on exchange 4 and "testroom" on exchange 5  
> > and one of
> > them fails? I wonder if the function should take a 'components' hash  
> > table instead of or
> > in addition to the name?
> 
> I started to write the API that way initially, actually, but 
> discarded  it as too complicated.  I think you're right that that 
> complication is  necessary; the AIM exchange example is a good one.
> 
> The easiest thing to do will be to keep the GHashTable used for chat 
>  creation around in the prpl's private data to be able to pass it  
> back.  Does that sound reasonable?

Yeah, that sounds like the best solution to me.

I like this new function, by the way.  I think it will be useful for meebo. 
Right now we look for Jabber chat join failures by doing string comparisons in
our notify UI code, which is a little ugly.

-Mark