| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
|
| |
Instead of creating a TpAccountManager through a factory, it is more natural
to create it directly. If no custom factory is provided at constructor,
a default one is created internally anyway.
This is also what tp-qt4 does.
Note this is break of unreleased API/ABI.
|
|
|
|
|
| |
We can now create TpSimpleHandler with account's factory. This avoid
introspection of other accounts if app does not need them.
|
|
|
|
|
|
|
|
|
| |
A factory is what we really need, forcing to create a TpAccountManager
also means that a simple channel handler will always introspect
all accounts even if it needs only one.
With a factory, TpBaseClient can create only needed accounts and still
share them with a TpAccountManager if one exists.
|
|
|
|
| |
This defines the TpAccountManager returned by tp_account_manager_dup()
|
|
|
|
| |
features are set on TpSimpleClientFactory now
|
|
|
|
|
|
|
| |
desired features
There are lots of code duplication for those proxy creation and preparation,
this will be fixed later.
|
|
|
|
| |
Use instead tp_simple_client_factory_ensure_account()
|
|
|
|
| |
prepared
|
| |
|
|
|
|
|
|
|
| |
accounts
For compatibility we still need to return a TpAccount and no ref,
so added a legacy table to keep them.
|
| |
|
| |
|
| |
|
|
|
|
|
|
| |
If dispose is run twice, first run would free the queue and 2nd would
still do a g_queue_foreach() on it... I can't beleive GLib still does
not have free_func and refcount on GQueue...
|
|\ |
|
| | |
|
| |
| |
| |
| | |
also means its features will be prepared on proxies
|
| |
| |
| |
| |
| |
| | |
contacts
contact-list-state must also be SUCCESS
|
| | |
|
| |
| |
| |
| |
| | |
If TpConnection has TP_CONNECTION_FEATURE_CONTACT_LIST, also create TpContact
objects for the roster. Add tp_connection_dup_contact_list() to get them.
|
|/ |
|
|\
| |
| |
| |
| | |
Fixes: <https://bugs.freedesktop.org/show_bug.cgi?id=26516>
Reviewed-by: Xavier Claessens <xclaesse@gmail.com>
|
| | |
|
| | |
|
| | |
|
| |
| |
| |
| |
| |
| | |
Fixes: fd.o#26516
Signed-off-by: Jonny Lamb <jonny.lamb@collabora.co.uk>
|
|\ \
| | |
| | |
| | |
| | |
| | | |
Fixes: <https://bugs.freedesktop.org/show_bug.cgi?id=27855>
Reviewed-by: Will Thompson <will.thompson@collabora.co.uk>
Reviewed-by: Xavier Claessens <xclaesse@gmail.com>
|
| | |
| | |
| | |
| | | |
We have a couple more synthesized properties these days.
|
| | |
| | |
| | |
| | |
| | | |
It is indeed guaranteed to be non-NULL because of the above: in fact,
the above set it to a non-NULL value in the handle != 0 case.
|
| | |
| | |
| | |
| | | |
Signed-off-by: Jonny Lamb <jonny.lamb@collabora.co.uk>
|
| | | |
|
| | |
| | |
| | |
| | | |
This is much nicer C API, and TpBaseClient had those helpers too.
|
| | | |
|
| | | |
|
| | | |
|
| | |
| | |
| | |
| | | |
the connection
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
TpAccountManager
Those are misleading, if app created its own TpSimpleClientFactory and is
using a TpAccountManager from it, they must pass their AM instead otherwise
the default one will be used.
|
| | | |
|
|\ \ \
| | | |
| | | |
| | | |
| | | | |
Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Fixes: https://bugs.freedesktop.org/show_bug.cgi?id=39666
|
| |/ / |
|
|/ /
| |
| |
| | |
It introspects Connection.ContactList properties
|
| |
| |
| |
| |
| | |
This is the same hypothetical issue as fixed (and tested to be fixed)
for SimplePresence.
|
| |
| |
| |
| |
| | |
Normally I'd say “use GQueue” but we can just as easily iterate
backwards.
|
| |
| |
| |
| |
| | |
You can't distinguish between “This contact has no information” and
“This contact's information is unknown” from the tp-glib API *anyway*.
|
| | |
|
| |
| |
| |
| |
| |
| |
| |
| | |
I was checking this code out and these scared me.
The one in tp_presence_mixin_set_status is particularly weird because
the callback was only called once because g_hash_table_foreach() only
got called if g_hash_table_size() returned 1!
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Previously, if a broken CM included the SimplePresence/presence
attribute when TpContact didn't ask for it,
CONTACT_FEATURE_FLAG_PRESENCE would still be set, even if we weren't
listening to PresencesChanged. So this is broken.
This fix makes TpContact ignore unsolicited presence information in
GetContactAttributes replies. The other option was to bind to the signal
in that event, which seems … surprising.
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
My reading of the spec says that it's okay to leave /client-types out of
the result of GetContactAttributes().
My reading of the tp-glib documentation is that it's valid for
tp_contact_get_client_types() to return NULL even if
TP_CONTACT_FEATURE_CLIENT_TYPES is prepared. TpContact:client-types is
documented a little differently, so I'm changing it to match.
|
| |
| |
| |
| |
| | |
The definition of the token attribute says it should be “omitted from
the result if the contact's avatar token is not known”.
|
| |
| |
| |
| |
| | |
The only CM that implements Location is Gabble. It implements Contacts.
Also Contacts has been around forever. Bye bye dead code.
|