| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
| |
https://bugs.freedesktop.org/show_bug.cgi?id=46430
|
| |
|
|
|
|
|
|
|
|
| |
In Automake 1.13, the long-deprecated macro AM_CONFIG_HEADER (deprecated
since 2002) has been removed in favour of AC_CONFIG_HEADERS.
Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=59604
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
tp_simple_client_factory_ensure_contact returns NULL for protocols, if
tp_connection_has_immortal_handles=0. Right now its happened for any
telepathy-python based protocols (butterfly, sunsine, mixer).
when tp_simple_client_factory_ensure_contact returns NULL,
tp_simple_client_factory_ensure_contact just segfault in
_tp_contact_set_attributes.
lets assume that contact sometimes NULL here, and skip them
Signed-off-by: Maksim Melnikau <maxposedon@gmail.com>
[edited for coding style, added a comment -smcv]
Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
Correctly get the remote contact handle out of the MediaDescription
hash, it's an a{sv} not string => uint32.
Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=54425
Conflicts:
telepathy-glib/base-media-call-content.c
|
|
|
|
|
|
|
|
|
|
| |
Remove the code that errors out a local description update if there is
*any* outstanding offer. A bit later in the same function a proper check
is done to see if there is a offer for the update contact, which is
correct.
Conflicts:
telepathy-glib/base-media-call-content.c
|
| |
|
|
|
|
|
|
| |
being NULL
https://bugs.freedesktop.org/show_bug.cgi?id=53201
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
| |
TpConnection itself wants to prepare TpContacts before being officially
CONNECTED. This was already the case for self contact. Use the same
trick for blocked contacts.
https://bugs.freedesktop.org/show_bug.cgi?id=52441
|
|
|
|
|
|
|
|
|
| |
We need to look for the .metadata file in the source directory,
otherwise we silently generate a binding with a namespace of "Tp", not
"TelepathyGLib".
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=49802
Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
|
| |
|
| |
|
|
|
|
|
|
|
|
|
| |
If we try to prepare a connection feature before the connection is connected,
the preparation may fall if a required interface hasn't been announced yet by
the CM. So, we give those features a second chance to be prepared when the
connection has been connected.
https://bugs.freedesktop.org/show_bug.cgi?id=42981
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
That function is dangerous because it does not guarantee that the returned TpContact
has an identifier set.
In the case a TpContact is being created with tp_connection_get_contacts_by_id(),
the first step is to inspect its handle and then create the TpContact object
*without* giving its ID. The id will be set later once we got the contact attributes.
If _tp_connection_lookup_contact() is called for the same handle/id between
the moment the TpContact is created and the moment it sets its attributes, then
a contact with no identifier is returned.
|
| |
|
|
|
|
| |
Reviewed-by: Jonny Lamb <jonny.lamb@collabora.co.uk>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Generated telepathy-glib/_gen/telepathy-enums-gtk-doc.h from:
/usr/bin/python ../tools/c-constants-gen.py Tp
_gen/stable-spec.xml _gen/telepathy-enums
has incorrect comment. Fix it with the same remedy as
in 5c58330c32f39a9dd6aaaef37f54461922cf6124
"many fixes to documentation comments"
ie remove the newline between '*' and identifier.
Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=48363
|
|
|
|
|
| |
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=48592
Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Instead of hard-coding the abstract Unix socket '\000dbus-tube-test',
listen on a random (abstract or real) Unix socket in /tmp if on Unix,
or a random port on 127.0.0.1 if not. These are the same listening
addresses I used in test/loopback.c in libdbus. The Unix version also
matches what the session bus normally uses.
This fixes the test on non-Linux Unix (not insisting on
abstract-namespace sockets working), when run twice in parallel on Linux
(not insisting on using a particular hard-coded name), and in theory
also on Windows (not trying to use Unix sockets).
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=48600
Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Reviewed-by: Guillaume Desmottes <guillaume.desmottes@collabora.co.uk>
|
| |
|
| |
|
|
|
|
|
| |
Since this is a stable branch, we're unlikely to fix uses of
deprecated functions, so we shouldn't warn about them.
|
| |
|
|
|
|
|
|
|
|
|
|
| |
* Build fails with new gtk-doc (1.18.1) which seems to
be stricter than previous versions.
[Also fixes g-ir-scanner warnings, which were what was actually fatal. -smcv]
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=48592
Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
| |
python-gobject can't bind G_TYPE_POINTER.
This seems to be save from an API/ABI pov according to GLib gurus,
we'll just copy the GError instead of passing the same pointer:
|
|
|
|
|
| |
This will allow us to use g_data_input_stream_read_line_utf8() in Python as
g_input_stream_read() is not bindable.
|
| |
|
|
|
|
| |
Make it easier to understand/debug what's going on.
|
|
|
|
|
|
| |
It is not always a good idea for bindings to always rely on GObject::get_property(),
it can be slow to access properties that way everytime instead of keeping
a proxy object in native language.
|
|
|
|
| |
Signed-off-by: Jonny Lamb <jonny.lamb@collabora.co.uk>
|
|
|
|
| |
Signed-off-by: Jonny Lamb <jonny.lamb@collabora.co.uk>
|
|
|
|
| |
Signed-off-by: Jonny Lamb <jonny.lamb@collabora.co.uk>
|
|\
| |
| |
| |
| |
| | |
https://bugs.freedesktop.org/show_bug.cgi?id=29271
Reviewed-by: Guillaume Desmottes <guillaume.desmottes@collabora.co.uk>
|
| | |
|
| | |
|
| | |
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Unlike with stream tubes, I don't think this is something which tp-glib
can decide for you. CREDENTIALS means “Use D-Bus's normal SASL
handshaking”; LOCALHOST means “turn off the same-user restriction and
optionally enable the ANONYMOUS mechanism if the socket really is only
available to local users” or something.
(To be honest, it's not obvious how an application can decide what it
wants without being patched. I think this functionality exists for
Sugar's benefit, where the CMs are run as different users to the
applications; the applications can only talk to them, IIRC, because the
session bus has authentication turned off.)
|