Problems using PurpleAccountConnect

Matthew Gordon matthew at
Wed May 13 17:53:53 EDT 2009

Thanks Etan. I've managed to fix all of my problems with your help. It's 
amazing to be able to control Pidgin using dbus.

-- Matthew

Etan Reisner wrote:
> On Fri, May 08, 2009 at 12:42:43AM +0200, Matthew Gordon wrote:
>> Hey Etan,
>> Thank you very much for the response.
>> What would I need to do differently to enable the account for Finch instead
>> of for Pidgin? It's more convenient for my usage to just use Finch, but I
>> can use Pidgin if necessary. When you refer to "enabling the account", are
>> you referring specifically to the call to SetEnabled? That invocation does
>> have some effect on Finch because I can check or uncheck the box next to
>> the account in Finch using SetEnabled. It seems a bit odd that it would do
>> that (and try to connect) but not actually put the account online.
>> -- Matthew
> Yes, "enabling the account" means calling purple_account_set_enabled.
> The fact that finch notices you calling purple_account_set_enabled with a
> ui string of "gtk-gaim" is a bug. The signal finch uses (account-enabled)
> doesn't pass the ui string along so finch needs to assume it was enabled
> for itself (because the majority of the time it will be enabling, and
> disabling, accounts for itself).
> This is a bug we should fix, but will need to wait for 3.0.0.[1]
> Until then you need to use the finch UI string, which you can find either
> by looking in the source, or by manually enabling an account in finch and
> seeing what it writes to the accounts.xml file[2].
> As to why the debug window says it is trying to connect but doesn't
> actually do anything, we print out that message before checking if the
> account is enabled, or if we have the right prpl loaded. (I'm fixing that
> as we speak, and adding a message for when someone tries to connect a
> disabled account.)
>     -Etan
> [1] Actually, we can fix the bug now by checking that the account was
> enabled for the current UI in purple_account_set_enabled. But the right
> fix would be to propagate the UI string to the signal so that UIs and
> plugins can get that information as well.
> [2] gnt-purple, I'm too nice. =)

More information about the Devel mailing list