| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
| |
https://bugs.freedesktop.org/show_bug.cgi?id=46430
|
| |
|
| |
|
|
|
|
|
|
| |
I started tracking this in Gabble so my unit test is living there.
https://bugs.freedesktop.org/show_bug.cgi?id=45982
|
| |
|
|
|
|
|
|
|
|
|
|
| |
Otherwise, if a caller kept a ref to the GAsyncResult after control had
returned to the channel, the channel could have freed the GString
already.
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=45554
Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Reviewed-by: Jonny Lamb <jonny.lamb@collabora.co.uk>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Without this change to TpAccount, the test would fail with a
use-after-free while inspecting reconnect_required. The TpAccount code
assumed that _finish would always be called directly from the callback,
but it is perfectly valid not to do so.
In the test, also test Reconnect (which is currently unimplemented),
since UpdateParameters and Reconnect are so closely related.
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=45554
Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Reviewed-by: Jonny Lamb <jonny.lamb@collabora.co.uk>
|
|
|
|
|
|
|
|
| |
lifetime
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=45554
Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Reviewed-by: Jonny Lamb <jonny.lamb@collabora.co.uk>
|
|
|
|
|
|
|
|
|
| |
This is valid usage, and often (as in this case!) uncovers bugs. It also
makes the flow of the code more obvious.
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=45554
Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Reviewed-by: Jonny Lamb <jonny.lamb@collabora.co.uk>
|
|
|
|
|
|
|
|
|
|
|
|
| |
It is valid to keep a GAsyncResult for as long as you like, so it will
have to take a copy of the result data. Otherwise, a change as simple as
replacing g_simple_async_result_complete with ..._complete_in_idle
will result in the data having already been freed by the time the
caller sees it.
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=45554
Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Reviewed-by: Jonny Lamb <jonny.lamb@collabora.co.uk>
|
|
|
|
|
|
| |
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=45554
Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Reviewed-by: Jonny Lamb <jonny.lamb@collabora.co.uk>
|
|
|
|
|
|
|
|
|
|
| |
If we emit Succeeded or Failed quickly enough after setting up the
TpProxy, the match rules might not have reached the dbus-daemon yet.
(The real ChannelRequest implementation avoids this bug by having the
Proceed method.)
Reviewed-by: Jonny Lamb <jonny.lamb@collabora.co.uk>
|
|
|
|
|
|
|
|
|
|
|
| |
If if a new contact fetch is queued from the callback of last
contact fetch, it leads to a crash. This is because when queueing
the new item it is processed right away since it is the only item
in queue, then when it returns from g_simple_async_result_complete()
it calls process_contacts_queue() again so the item is prepared twice,
and then freed twice.
Fix this by keeping currently being processes item outside the queue.
|
|
|
|
|
|
| |
If no account is connected, doc says we should return
(TP_CONNECTION_PRESENCE_TYPE_OFFLINE, "offline", "") but we use to return
(TP_CONNECTION_PRESENCE_TYPE_OFFLINE, NULL, NULL)
|
| |
|
| |
|
| |
|
|
|
|
|
|
| |
An unknown owner is implemented storing a NULL TpContact in the hash table.
https://bugs.freedesktop.org/show_bug.cgi?id=43755
|
|
|
|
|
| |
Convention is to use the _async method as that's what being checked in
tp_account_set_uri_scheme_association_finish().
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
| |
Sjoerd tells me that “with older g-i you don't get all annotation right
and gnome-shell will fail”. Specifically, methods like
DelegateChannel(), used by Gnome Shell 3.2, don't work. 1.30 is the
current stable release.
|
|
|
|
|
|
|
|
|
|
|
|
| |
Vala-0.12 does not generate the constructor for ContactInfoField,
which causes a build failure for Folks (and any other clients
that depend upon creating new ContactInfoFields from Vala).
Empathy 3.2, for whose benefit the 0.16 stable branch of tp-glib exists,
depends (via Folks) on telepathy-glib being built with at least
valac-0.14.
https://bugs.freedesktop.org/show_bug.cgi?id=42296
|
|
|
|
|
| |
The pointer goes stale when the result is consumed,
and it trips an assertion if the manager gets a new challenge.
|
| |
|
| |
|
|
|
|
|
|
|
| |
This is the case when joining a room containing members having an unknown
owner.
https://bugs.freedesktop.org/show_bug.cgi?id=42670
|
|
|
|
|
| |
g_return_if_fail() can be compiled as noop, so we shouldn't rely on it to get
the handles.
|
|
|
|
|
|
|
| |
We unconditionally implements the MembersChangedDetailed signal so this flag
should always be set.
https://bugs.freedesktop.org/show_bug.cgi?id=42305
|
|
|
|
|
| |
It turns out that when we talked about invalidation, people didn't actually
know what this meant.
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
| |
As part of preparing a Connection the self contact will be retrieved. If
the contacts requested features include _CAPABILITIES _and_ the
connection doesn't support ContactCapabilities then the Capabilties of
the connection get use. But if we wait for this feature to be prepared
then the introspection stalls.
So instead use internal API to force getting the capabilities of the
connection regardless of the state of the various features and break the
circular dependency.
|
| |
|
|
|
|
|
|
|
| |
If we don't implement Requests we shouldn't respond to the retrieval of
RCC properties. Unfortunately that's tricky to do, so do a quick hack
which means we will have an empty RCC property, which is at least
somewhat more useful for our tests.
|
|
|
|
| |
some cases
|
| |
|
|
|
|
|
|
| |
TP_TOKEN_CONNECTION_INTERFACE_CONTACT_BLOCKING_BLOCKED
https://bugs.freedesktop.org/show_bug.cgi?id=42049
|
|
|
|
|
|
|
|
| |
If the owner is unknown self->priv->group_contact_owners may contain NULL
TpContact. g_object_unref() not being NULL safe we have to use our own
value_destroy_func to avoid warnings.
https://bugs.freedesktop.org/show_bug.cgi?id=41928
|
| |
|
|
|
|
|
|
|
|
|
| |
not NULL
For example, when receiving a MUC delivery report we end up with a message
having no sender and so no contact to prepare.
https://bugs.freedesktop.org/show_bug.cgi?id=41929
|
| |
|
| |
|
| |
|
| |
|
|
|
|
| |
Fixes: <https://bugs.freedesktop.org/show_bug.cgi?id=41714>
|
|
|
|
|
|
|
|
|
| |
This macro is really ugly. I only left it in place for assertions like
MYASSERT (!tp_proxy_prepare_finish (chan, prepare_result, &error), "");
where the assertion has a side-effect. Otherwise, if someone disables
assertions the test will crash. Ugh.
|
|
|
|
| |
fea8294 introduced this pretty obvious leak.
|