summaryrefslogtreecommitdiff
AgeCommit message (Collapse)AuthorFilesLines
2014-04-01Restore lost test coverage from get-interface-after-invalidate testnext-factorySimon McVittie1-1/+22
2014-04-01TpProxyClass: disobeying must_have_unique_name is programming errorSimon McVittie1-2/+3
2014-04-01Add channel-filter to documentationSimon McVittie1-0/+1
2014-04-01Rename remaining tp_dbus_daemon_*() to tp_dbus_connection_*()Xavier Claessens48-662/+589
Also moves their code to dbus.c/h module since their are only GDBusConnection helpers now, not methods on TpDBusDaemon object.
2014-04-01Remove tp_dbus_daemon_get_unique_name()Xavier Claessens9-27/+5
2014-04-01Remove TpDBusDaemon objectXavier Claessens16-363/+7
2014-04-01Replace all occurences of TpDBusDaemon by GDBusConnectionXavier Claessens100-784/+480
This is a huge commit but almost brainless. There is a notable change in tp_tests_proxy_run_until_dbus_queue_processed() to support taking a GDBusConnection instead of a proxy.
2014-04-01Stop passing a TpDBusDaemon to TpProxy and subclassesXavier Claessens23-34/+4
2014-04-01TpClientFactory: Fix documentation for custom singletonXavier Claessens1-4/+4
2014-04-01Rename _new_with_factory() to simply _new()Xavier Claessens21-127/+61
Proxies are not always created with a factory so it's not a special constructor anymore. This also removes the TpDBusDaemon argument because TpProxy::constructed will fill it itself from the factory anyway.
2014-04-01TpProxy: Assert that we have a factoryXavier Claessens1-28/+27
We can also deduct the TpDBusDaemon and GDBusConnection from the factory.
2014-04-01Tests: Pass a factory to all TpProxy instancesXavier Claessens11-32/+80
Lots of tests creates plain TpProxy instances, they need a factory as well because TpProxy::constructed will soon start to assert that it got a factory.
2014-04-01Farstream: Propagate the call channel's factory to other TpProxy instancesXavier Claessens2-0/+2
2014-04-01TpProxy: Remove now unused _tp_proxy_ensure_factory()Xavier Claessens2-24/+0
2014-04-01Use constructed instead of constructor in TpProxy and subclassesXavier Claessens4-66/+41
2014-04-01tp_client_factory_dup_logger: rename to _ensure_loggerSimon McVittie4-5/+5
2014-04-01TpLogger: Always create from a factoryXavier Claessens5-27/+54
2014-04-01TpConnectionManager: Always create from a factoryXavier Claessens14-79/+145
Also create CM's TpProtocol objects from the factory now that the CM always has one. In protocol-objects test_factory() it changes the cm_name because TpProtocol objects are now shared with TpConnectionManager that keeps a ref on them.
2014-04-01TpConnectionManager: Take a factory to list CMsXavier Claessens6-20/+31
The factory will be used to create TpConnectionManager objects in the next commit
2014-04-01TpDebusClient: Always create from a factoryXavier Claessens8-25/+53
2014-04-01tp_client_factory_dup_channel_dispatcher: rename to _ensure_channel_dispatcherSimon McVittie7-9/+9
2014-04-01TpChannelDispatcher: Always create from a factoryXavier Claessens11-28/+49
Conflicts: docs/reference/telepathy-glib/telepathy-glib-sections.txt telepathy-glib/client-factory.h
2014-04-01tp_client_factory_dup_account_manager: rename to _ensure_account_managerSimon McVittie5-6/+7
Xavier was going for "singletons use _dup_, objects that are per-path use _ensure_" but we agreed that using _ensure_ for both was less confusing.
2014-03-29TpAccountManager: Always create from a factoryXavier Claessens9-171/+67
2014-03-29TpClientFactory: Add getter for its GDBusConnectionXavier Claessens3-0/+21
2014-03-29TpClientFactory: Add a singleton factoryXavier Claessens3-0/+107
tp_client_factory_dup()'s documentation is a preview from what it will be in a future commit. I did not write a temporary documentation in the meantime.
2014-03-29TpClientFactory: Simplify TpProtocol codeXavier Claessens3-44/+32
2014-03-27Logger: read the right schemanext-after-gdbusSimon McVittie1-1/+1
2014-03-27telepathy-logger: rename the daemon to a versioned nameSimon McVittie3-4/+6
2014-03-27Logger: rename schema to be parallel-installableSimon McVittie3-3/+3
The schema ID already didn't correspond to the path.
2014-03-27tp_run_connection_manager: disable exit-on-close for CMsSimon McVittie1-0/+3
2014-03-27Add a comment to clarify how tp_dbus_daemon_unregister_object() worksSimon McVittie1-0/+2
2014-03-27tp_dbus_daemon_try_register_object: avoid GSListSimon McVittie1-4/+4
The space optimization is hardly relevant here. "I think GSList's only purpose is making your app crash if you use g_list_free on it" -xclaesse
2014-03-27tp_proxy_get_interface_by_id: removeSimon McVittie5-74/+10
2014-03-27TpDBusDaemon: Remove code for listing namesXavier Claessens4-108/+0
The few places where it was used can use GDBusConnection directly Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
2014-03-27Tests: Use GDBusConnection directly to list activatable namesXavier Claessens1-8/+20
Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
2014-03-27TpConnectionManager: Use GDBusConnection directly to call ListNamesXavier Claessens1-253/+120
This rewrites tp_list_connection_managers_async() to simplify the implementation. Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
2014-03-27TpConnection: Remove tp_list_connection_names()Xavier Claessens6-241/+0
Applications should use TpAccountManager to list online accounts. The only user of this function is MissionControl but it can call ListNames itself. Note that example code was even wrong. It says "list installed connection managers" but what is actually do is listing currently running CMs. Conflicts: examples/client/Makefile.am Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
2014-03-27TpDBusDaemon: Remove code for watching namesXavier Claessens5-617/+1
GDBusConnection provide similar feature already. Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
2014-03-27TpBaseConnection: Re-implement client interest using GDBusConnectionXavier Claessens1-161/+126
Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
2014-03-27Tests: Use GDBusConnection directly to watch nameXavier Claessens1-5/+9
[explicitly cancel the watch -smcv] Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
2014-03-27TpDebugClient: Use GDBusConnection directly to watch nameXavier Claessens1-24/+22
Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
2014-03-27TpConnectionManager: Use GDBusConnection directly to watch nameXavier Claessens1-68/+78
Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
2014-03-27TpAccountManager: use GDBusConnection directly to watch nameXavier Claessens1-19/+17
Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
2014-03-26Debug sender: avoid using codegen for signal emissionnext-gdbusSimon McVittie1-2/+37
At the moment telepathy-glib has a lot of g_debug() while we're debugging the new GDBus-based implementations, so the codegen's signal emissions do a g_debug(), which triggers a g_debug(), and so on.
2014-03-26TpProxy: strip detailed error name from error->message if domain is knownSimon McVittie1-0/+4
Some of the Mission Control tests assert that particular error messages are passed-through nicely, which caught this.
2014-03-26name owner watching: more debug, and use a free-function to free watchesSimon McVittie1-13/+20
Several Mission Control tests crash when trying to add more callbacks to a _NameOwnerWatch that has not been removed from the hash table after calling _tp_dbus_daemon_stop_watching(). It's not clear to me how that can happen, but turning _tp_dbus_daemon_stop_watching() into an ordinary free-function ensures that "remove it from the hash table" and "free it" happen in the right order.
2014-03-26Update ABI files for GDBus branch, and sort with LC_ALL=CSimon McVittie3-85/+193
The majority of the changes are because the new codegen generates extern symbols for the _return_from_ family of functions.
2014-03-26document TpProxyWrapperFuncSimon McVittie1-0/+16
2014-03-26Document tp_proxy_pending_call_v1_newSimon McVittie1-0/+30
generated by cgit v1.2.3 (git 2.39.1) at 2024-11-24 15:18:38 +0000