[Telepathy] ANNOUNCE: telepathy-idle 0.1.3 released
Jonathon Jongsma
jonathon.jongsma at collabora.co.uk
Wed Feb 18 09:04:45 PST 2009
Telepathy-idle is a connection manager for the IRC.
This is a bugfix release. It fixes several long-standing bugs (including some that lead
to loss of messages), and fixes several significant memory leaks. Users of previous
releases are highly encouraged to upgrade to this release. It can be downloaded from the
following location:
Tarball:
http://telepathy.freedesktop.org/releases/telepathy-idle/telepathy-idle-0.1.3.tar.gz
Signature:
http://telepathy.freedesktop.org/releases/telepathy-idle/telepathy-idle-0.1.3.tar.gz.asc
Changes since 0.1.2
=================================
Dependencies:
* telepathy-glib 0.7.15
Enhancements:
* Added automated test infrastructure
Fixes:
* Fixed a bunch of memory leaks reported by valgrind
* handle buggy servers that don't quit properly
* Fix parsing error that resulting in not receiving any messages that started
with a leading space (Bug #17390)
* Handle user renames in room channels properly
* Validate nicknames properly when attempting to create a new connection
* Fix bug where data was lost due to improper splitting of long messages
(Bug #17392)
* Don't split utf-8 codepoints in half when splitting long messages (Bug #13532)
* Fix bug where a user named 'foo' that was a member of a channel of the same
name ('#foo') would have private messages delivered to both the private chat
and the group chat (Bug #19766)
* Fix some infinite loop failures when connections fail
* Fixed various warnings
Full changelog follows:
commit ceaaab52fee088514eb79dba0bf133ee68d1e281
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Wed Feb 18 10:32:18 2009 -0600
version 0.1.3
commit 510dc386b32adc28fc8309c538c083c413b9e09b
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Wed Feb 18 10:15:52 2009 -0600
Fix builddir!=srcdir issues
Some of the tests were not running properly in distcheck since I had forgotten a
couple of EXTRA_DIST incantations and was trying to load the ssl cert/key files
from the current directory rather than the src directory. The latter issue I
worked around by setting an env variable for the location of the files since I
really didn't want to do something like idletest.py.in to get autoconf to
pre-process the python source file to get the correct srcdir location
commit 4fb27c730a15b5d37ba6ce9769c71d720896af1b
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Tue Feb 17 23:31:03 2009 -0600
Update the text-encode-and-split test to the new internal API
I had added a 'max lenth' parameter to the function but apparently forgot to
update this test.
commit 9807d0f50e6752f958c1ee071aaeeb1cf536f995
Merge: 619cc9d... dbdbbb9...
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Tue Feb 17 17:02:30 2009 -0600
Merge branch 'long-message-split'
Conflicts:
tests/twisted/Makefile.am
commit 619cc9d78913d8f0a0d48e5e52aeb404e46884c6
Merge: 98c90f3... 8cb7eba...
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Tue Feb 17 15:24:30 2009 -0600
Merge branch 'twisted-tests'
commit 98c90f34fab760c8207ae0101b79a3f8b6dff7fc
Merge: 6933765... 57eff23...
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Tue Feb 17 15:24:20 2009 -0600
Merge branch 'invalid-nick'
Conflicts:
tests/twisted/Makefile.am
commit 8cb7ebaddd424c08a686edf5b46f71125b3a1945
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Tue Feb 17 14:58:32 2009 -0600
Fix a test crash in connect/server-quit-ignore.py
This test was only failing when I ran the full test suite, not when I ran the
test by itself. Running under valgrind indicated that my _force_disconnect()
timeout function was attempting to use an already-freed object:
==19693== Invalid read of size 8
==19693== at 0x52B02AC: g_type_check_instance_cast (in
/usr/lib/libgobject-2.0.so.0.1800.4)
==19693== by 0x407719: _force_disconnect (idle-connection.c:438)
==19693== by 0x55042BA: (within /usr/lib/libglib-2.0.so.0.1800.4)
==19693== by 0x5503AFA: g_main_context_dispatch (in
/usr/lib/libglib-2.0.so.0.1800.4)
==19693== by 0x55072BC: (within /usr/lib/libglib-2.0.so.0.1800.4)
==19693== by 0x55077EC: g_main_loop_run (in /usr/lib/libglib-2.0.so.0.1800.4)
==19693== by 0x5E0A9C5: tp_run_connection_manager (run.c:264)
==19693== by 0x406504: main (idle.c:38)
==19693== Address 0x6cd3258 is 0 bytes inside a block of size 328 free'd
==19693== at 0x4C2161F: free (vg_replace_malloc.c:323)
==19693== by 0x52B17BA: g_type_free_instance (in
/usr/lib/libgobject-2.0.so.0.1800.4)
==19693== by 0x52B45C2: g_value_unset (in /usr/lib/libgobject-2.0.so.0.1800.4)
==19693== by 0x52A7207: g_signal_emit_valist (in
/usr/lib/libgobject-2.0.so.0.1800.4)
==19693== by 0x52A76D2: g_signal_emit (in /usr/lib/libgobject-2.0.so.0.1800.4)
==19693== by 0x4076F5: _finish_shutdown_idle_func (idle-connection.c:430)
==19693== by 0x5503AFA: g_main_context_dispatch (in
/usr/lib/libglib-2.0.so.0.1800.4)
==19693== by 0x55072BC: (within /usr/lib/libglib-2.0.so.0.1800.4)
==19693== by 0x55077EC: g_main_loop_run (in /usr/lib/libglib-2.0.so.0.1800.4)
==19693== by 0x5E0A9C5: tp_run_connection_manager (run.c:264)
==19693== by 0x406504: main (idle.c:38)
GLib-GObject-WARNING **: invalid unclassed pointer in cast to `IdleConnection'
aborting...
Cancelling the timeout when shutting down the connection seems to fix it. Now
all tests pass again \o/
commit 7ccb8c5ddc04ecaaec12b5a85ccc7a0d828269a9
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Mon Feb 16 22:16:29 2009 -0600
Fix a small leak reported by valgrind
Valgrind report was:
==24778== 240 bytes in 2 blocks are definitely lost in loss record 22 of 33
==24778== at 0x4C2291E: malloc (vg_replace_malloc.c:207)
==24778== by 0x550C272: g_malloc (in /usr/lib/libglib-2.0.so.0.1800.4)
==24778== by 0x5539314: g_io_channel_unix_new (in /usr/lib/libglib-2.0.so.0.1800.4)
==24778== by 0x4100E5: ssl_do_connect (idle-ssl-server-connection.c:480)
==24778== by 0x4101E6: ssl_dns_result_cb (idle-ssl-server-connection.c:504)
==24778== by 0x41136D: _resolve_idle_func (idle-dns-resolver.c:154)
==24778== by 0x5503AFA: g_main_context_dispatch (in /usr/lib/libglib-2.0.so.0.1800.4)
==24778== by 0x55072BC: (within /usr/lib/libglib-2.0.so.0.1800.4)
==24778== by 0x55077EC: g_main_loop_run (in /usr/lib/libglib-2.0.so.0.1800.4)
==24778== by 0x5E0A9C5: tp_run_connection_manager (in
/usr/lib/libtelepathy-glib.so.0.22.1)
==24778== by 0x406504: main (idle.c:38)
This was observed while valgrind-ing the connect-ssl-fail.py test
commit 9fcc28e7ad38f40c105aaff735b284b738e921a0
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Mon Feb 16 21:32:36 2009 -0600
Fix another infinite loop on connection failure
This is similar to the ssl infinite loop that was fixed earlier. I'm not sure
if this was introduced in a faulty merge or if I somehow overlooked it before :/
Uncovered by the connect/connect-fail.py test
commit 36348b1fae8f1ed454183aa1be93005fde64beaa
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Tue Feb 17 12:06:27 2009 -0600
messages/leading-space.py: simplify this test significantly
wjt questioned the need for the subclassed server in his review and eliminating
the server actually makes things much simpler. At the same time, I realized
that there's really no need for the user to join a chat room in order to test
out this behavior, so remove that whole piece and just send a privmsg directly
and test whether it is parsed correctly.
commit 44e4d1f6e6373213a52a1f5c01b86fa7fe3bdadb
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Tue Feb 17 11:19:44 2009 -0600
Rework join-muc-channel.py test to not depend on object path
Yes, this is much much cleaner. Also cleaned it up a lot by using constants
from constants.py
commit 01fd11cb43fc8890bd3eba69c639a7eeba268dae
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Tue Feb 17 10:53:27 2009 -0600
message-order test: use loop to make this a bit cleaner
commit 44ec4723a51a3d9aa9fe0ccb8530a1c91522f147
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Mon Feb 16 22:57:42 2009 -0600
Disconnect in idletest in case the test fails and doesn't disconnect
This theoretically should prevent a failing test to affect the following tests
Also remove a stray mention of gabble. Findings from wjt's review
commit fccb2e9f055b0a539174b4e7f425f1beb93d6442
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Mon Feb 16 22:53:14 2009 -0600
Reduce test chatter when not in verbose mode
I was print()ing a lot of things in my BaseIRCServer class, which was cluttering
up the test output and making it hard to easily glance and see the results.
Modified so that it only prints out when in verbose mode.
commit c3649eaad00b88b487d491ce119d65ebcfb01627
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Mon Feb 16 22:39:42 2009 -0600
Add the constants.py file from gabble
Use the constants in a few tests
commit 57eff2309968c8ac6157194d5183bf384e823db5
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Fri Feb 13 23:52:16 2009 -0600
Clean up some connections in invalid-nick test
Also assert that the dbus exception is the correct one. Changes per wjt's
review.
commit dbdbbb96349443a48e741f07d175808899bc343f
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Fri Feb 13 23:20:35 2009 -0600
Assume worst-case relay prefix before we get user info
Finding from wjt's review. I wasn't able to find a max limit on userlength, but
the 3 IRC servers that I tested (freenode, gimpnet, and ircd-hybrid on localhost
as shipped by debian) all had limits between 8-10 characters. I added a little
padding and made username worst-case 18 chars.
commit 6933765064960d84c71747873851393ed88f4c9b
Merge: ad2a004... d471f42...
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Fri Feb 13 19:39:19 2009 -0600
Merge branch 'room-contact-mixup'
commit ad2a004f8aa490f756708c42b0d8b1f1759584c1
Merge: fda6247... 4a64d98...
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Fri Feb 13 19:37:25 2009 -0600
Merge branch 'gimpnet-quit'
Conflicts:
src/idle-server-connection.c
commit fda62472258ddbb0cb512de5958763ada028e5d1
Merge: 39ee883... ae92668...
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Fri Feb 13 17:38:49 2009 -0600
Merge branch 'leakfix'
commit 39ee883cfe3ede415fee8f4ae87636cc2024053c
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Thu Jan 22 11:29:25 2009 -0600
Add documentation for IRC message spec
If you haven't worked with it in a little while, it can be a little challenging
to remember what the different characters mean for the IRC message spec. This
just adds a small comment describing each of the format letters so that it's
easier to add new IRC message types in the future.
commit efe60c1b62ebf41c2c2f0336d7b5322454744a92
Merge: 18a3cd1... 492c891...
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Fri Feb 13 17:31:52 2009 -0600
Merge branch 'fd-double-close'
Conflicts:
src/idle-ssl-server-connection.c
commit 18a3cd12be7976cb00123537b2575d572feedf61
Merge: 7730041... 40fa184...
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Fri Feb 13 17:30:24 2009 -0600
Merge branch 'ssl-infinite-loop'
commit d471f42675222cce620a219b36791e7f0e69bb5f
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Wed Jan 28 11:27:19 2009 -0600
Re-work the room-contact-mixup test with alban's suggestions
use a signal handler on the dbus signal to set a global flag and assert that the
flag is false at the end rather than waiting for a signal we don't want to get.
I tested that this still failed without the patch and passes after, and it makes
the test a bit cleaner (and runs faster).
commit a88b5955d95ba5ed1fa7cd23b574883c5a017543
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Tue Jan 27 17:24:59 2009 -0600
Fix Bug #19766: private messages no longer delivered to group chat
This could perhaps be a slightly controversial change. The reason that the bug
happened was that telepathy-idle's room handle repository normalizer function
had some special code to make joining chatrooms more user-friendly. Basically,
if the user specified a chatroom 'foo', it would assume you meant '#foo' and
automatically prepend the '#' character.
The way the parser currently works is that when we get in a privmsg, we try to
parse the 'destination' of the privmsg as both a contact and a room. The way we
determine if it's a valid room or not is by calling tp_handle_ensure(room_repo,
...) (and same for determining valid contacts). Unfortunately, because of the
automatic #-prepending this means that a single privmsg destination can be both
a valid contact and a valid channel name, so if the user is in a channel with
the same name as their nick (but without the leading #), the privmsg will be
handled by both the IMFactory and the MUCFactory
commit 919ee9b1448430737f89056209280947ec5ade72
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Tue Jan 27 17:23:12 2009 -0600
Add test for Bug #19766: private messages deliver to group chat
This test currently fails because the user joins a channel with the same name as
the user, and when a remote contact sends a private message to the user it gets
signalled as 'Received' on both the private chat channel and the group chat
channel
commit 155488bc72498bb1c2acf78cd4a001836e21e54b
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Fri Feb 13 16:05:51 2009 -0600
Split long messages properly (Bug #17392)
Use the newly-added idle_connection_get_max_message_length() API to split long
messages so that they allow for the server to add a prefix onto the messages
that it relays to others instead of simply using 512 as the max message length
I had to modify the test a bit to get it to behave like a real IRC server, but
now the test passes fine, and I did quite a bit of manual testing (copy/pasting
large chunks of test between two accounts) and have not seen any issues yet.
Part of my manual testing included copy/pasting large amounts of unicode text
from various wikipedia articles in other languages, so I believe that the utf-8
splitting issue is also fixed properly in this branch.
commit 5edc8f54cafe32ac1daf37a9c940ee81b2ce92f8
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Fri Feb 13 15:20:43 2009 -0600
Store our user host prefix to calculate max msg length
After logging in, query information about ourselves so that we know what prefix
string will be tacked onto the beginning of our messages by the server when they
are relayed on to other users. This will allow us to calculate our max message
length properly.
This patch also adds a new API method idle_connection_get_max_message_length()
that allows others to query this information without knowing anything about the
actual host / prefix string.
commit c573b2b2cfffd13062dee546a150798ec4356cb3
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Fri Feb 13 13:54:36 2009 -0600
Don't split a utf-8 codepoint in half when splitting long message
Basic approach: find the place we would normally break the message, and walk
back to the previous utf-8 character start.
commit 2c8cce859dafdd85d409010c505883a6c24da8a7
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Fri Jan 30 15:47:40 2009 -0600
Add a test for improper splitting of long messages (Bug #17392)
See the bug for details, but essentially, telepathy-idle will internally split
long messages (those above the IRC max length of 512) and send them in multiple
messages. The problem is that the server then has to relay them on to the other
members of the channel, and when it does this, it adds a ':nick!server.name here
' prefix onto the message. So if our message was exactly 512 bytes, when this
prefix gets added, it needs to chop some of the message off the end to fit under
the 512-char limit. It does so silently, unfortunately. So we'll need to
either limit our max messages to a value considerably lower than 512 to leave
room for the server to add their prefix or we'll need to figure out a way to
calculate very closely what the prefix is going to be (for the longest possible
nickname)
commit dfb5b76c455830428d6481e6318b184a456c2b46
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Fri Jan 30 14:51:37 2009 -0600
Validate nicks per the IRC RFC
Previously we were allowing the nick to consist of any unicode alpha or digit
characters, but the IRC spec (RFC2812 section 2.3.1) is very explicit about
which characters are valid for a nick, and it's only ascii chars.
In addition, we weren't disallowing '-' as the first character of a nickname as
required by the IRC spec.
Added a few additional tests for valid and invalid nicks
commit 00ead0bb95daa75de7be57f555f5fa5f05371cb3
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Fri Jan 30 13:45:25 2009 -0600
Expose the nick validation function and use it in the CM
Add a filter for the 'account' param of the RequestConnection method so that it
can check if the passed nickname is valid for an IRC nick and reject it if not.
Previously, telepathy-idle happily accepted invalid nicknames and then didn't
provide any decent feedback about what was wrong when the 'login' failed.
commit ac268e98767da9a2e4d10155fa612cb5b55bb8ef
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Fri Jan 30 13:43:56 2009 -0600
Add test for invalid nicknames
telepathy-idle should not blindly accept invalid nicknames -- it should do a
validity check on them and reject them if they're not valid IRC nicks. This
test does that. Fix will come soon.
commit 6d197ed30d0f4fb17d316be966206782cbbfe1e3
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Thu Jan 29 12:24:27 2009 -0600
Speed up the tests a bit
When I was figuring out the test framework and trying to get things to work, I
had bumped up the time out on a bunch of tests to 10s rather than the default 5.
This means that a timeout failure takes a lot longer to happen and there's
really no need for a full 10s timeout. Also, when testing that messages are
sent in the right order, there's no need to send 10 messages, so cut a few of
them. Since there's a built-in 2s delay between each sent message, that means
the test was always taking *at least* 20s to execute.
commit 4a64d982e41406d7ca738a8025d70566ca7cbbf8
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Tue Jan 27 14:52:21 2009 -0600
Handle buggy servers that don't quit properly
This fix kills 2 related bugs at once (and thus fixes both of the tests that I
just added on this branch).
1. It disables receiving further messages after we send our QUIT message, so
that we don't process a stray "welcome" message and accidentally try to
transition to CONNECTED state while we're in the middle of disconnecting.
2. Schedules a forceful server disconnection for 2s after we send our QUIT
message in the case that the remote server does not close the socket. This
means that we send the dbus method return for Disconnect() at most 2s after
receiving it, rather than depending on the remote server to disconnect us
before we send the dbus method return. I suppose I could have just
disconnected immediately rather than scheduling it for 2s in the future, but
I wanted to stay as close to current behavior as possible.
commit 438f8edc058ae17998c443c80e01892658eb93fb
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Tue Jan 27 13:52:08 2009 -0600
Add a couple tests for slightly buggy server behavior
gimpnet IRC server has a bug where if you QUIT while it's waiting for an ident
response (at 'login'), it will totally ignore the QUIT message and go on to send
you a welcome message. I've added a test which approximates this behavior.
In addition, if you send quit behavior, but the remote server doesn't close the
socket, telepathy-idle will never actually quit (or even respond to the
Disconnect() dbus method, in fact). So I've added a test for this case.
I think we should probably unilaterally close the connection and exit after a
certain time if the server is misbehaving in one of these ways
commit a6e409abbcd7591e7fd8689b17f1b0074cbad624
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Tue Jan 27 12:11:58 2009 -0600
Make the test server respond to QUIT messages properly
Previously I was just closing the connection immediately upon receiving a QUIT
message, but I should have been sending an ERROR response and then closing the
connection.
commit ae926684ebf091e3ecc4dd29761563bb30c5fc14
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Mon Jan 26 15:15:23 2009 -0600
Free a bunch more memory leaks reported by valgrind
Many telepathy signals take GPtrArrays of specialized dbus-glib boxed types.
The basic pattern for all of these leaks is the same:
- create a GValue for the specialized dbus type
- call g_value_take_boxed(foo, specialized-dbus-type);
- dbus_g_type_struct_set(...) /* set struct props */
- g_ptr_array_add(array, g_value_get_boxed())
- emit tp signal
- free ptr array
The problem is that even though the pointer array of the boxed types are freed,
the boxed types themselves are not properly freed. So you need to go through
the list and call g_boxed_free on all of the array elements before freeing the
array (or call g_value_unset() after you're done with the ptr array, which will
result in the boxed type being freed by the GValue since it was initialized with
take_boxed()).
In addition, there was a somewhat related/intertwined bug in the topic handlers:
- in idle_muc_channel_topic_touch(), where we were calling g_value_get_boxed()
and adding it to the GPtrArray, then modifying the GValue, calling
g_value_get_boxed(), and adding that to the array again. The intent was to
add two different objects to the array, but since we were not copying the
boxed type (e.g. g_value_dup_boxed()), the result was that we were adding a
pointer to a boxed type in the array, modifying the object and then adding it
again, so we ended up with two identical boxed type objects in the array
rather than two different objects. I'm not entirely sure what effect this bug
had on things, but it should be fixed now regardless.
- in idle_muc_channel_topic_full(), we had the same problem as above, but in
addition, the 'subject' property was never actually added to the ptr array, so
I added this call.
commit 10edb283b49d35863cb2345b3fe308ba5ad40180
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Mon Jan 26 10:15:48 2009 -0600
Fix a leak in change_tp_properties()
We were leaking all GValue* objects that we got with dbus_g_type_struct_get()
every time we called change_tp_properties(). This valgrind error is
representative of the leaks:
==2405== 1,348 (280 direct, 1,068 indirect) bytes in 14 blocks are definitely lost in
loss record 5 of 15
==2405== at 0x4023DE2: calloc (vg_replace_malloc.c:397)
==2405== by 0x4389479: g_malloc0 (gmem.c:151)
==2405== by 0x43060A1: value_copy (gboxed.c:94)
==2405== by 0x43069CE: boxed_proxy_lcopy_value (gboxed.c:362)
==2405== by 0x42B08C5: dbus_g_type_struct_get (dbus-gtype-specialized.c:978)
==2405== by 0x805D1EC: change_tp_properties (idle-muc-channel.c:526)
==2405== by 0x805F39D: idle_muc_channel_topic (idle-muc-channel.c:1222)
==2405== by 0x8053969: _numeric_topic_handler (idle-muc-factory.c:184)
==2405== by 0x8055F5D: _parse_and_forward_one (idle-parser.c:399)
==2405== by 0x8055BC7: _parse_message (idle-parser.c:321)
==2405== by 0x8055694: idle_parser_receive (idle-parser.c:214)
==2405== by 0x804EFF6: sconn_received_cb (idle-connection.c:656)
After this commit, this particular error message no longer appears, though there
is one additional leak that I need to chase down yet.
commit 8c87b6e3b219d720038420201382b1b66ab7e789
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Mon Jan 19 16:14:02 2009 -0600
Handle connect()s that return success immediately
The documentation on connect() is cryptic about in what situations a
non-blocking socket can connect immediately, but in theory it is possible. I
had several test failures caused by violating the assumption that the return
status from connect() is always -1 on a non-blocking socket. These turned out
to be caused by attempting to connect a UDP socket (which apparently is
instantaneous even on non-blocking sockets), but it seems wise to handle this
case more gracefully anyway, rather than simply asserting.
commit 7730041cb68402b8eadafc57d767cf8626638a00
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Mon Jan 19 15:48:41 2009 -0600
Only try to connect to the IRC server over TCP
Previously we were not specifying any socket type in our getaddrinfo call, so
our address results included both TCP and UDP candidates. When connecting to
the TCP one failed, it would try the UDP one, which would lead to some
unexpected behavior (for instance, connect() on a UDP socket with O_NONBLOCK set
will return 0 rather than -1/EINPROGRESS, which violates some assumptions in the
code). If the UDP socket did manage to get connected, it would violate other
assumptions (for instance, setting the TCP_NODELAY option doesn't make sense on
a non-TCP socket and triggers a warning).
So, it doesn't really make sense to connect to a IRC server with UDP anyway, so
this patch restricts telepathy-idle to use TCP sockets only
commit 40fa1847606e4176432e2177a55a87934450a16b
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Mon Jan 19 15:45:47 2009 -0600
Fix an infinite loop in the SslServerConnection class
This failure was triggered by the connect-fail-ssl test. When a ssl connection
fails, it just retries the exact same address over an over in an infinite loop.
It doesn't move on to the next address as it should.
commit fbd509b4e34f0e96fbd2a87e18fbecfdfedaef70
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Mon Jan 19 15:44:48 2009 -0600
Add a little explanation to the connect-fail-ssl test
commit d0e9f8ddcb99f3d7b848be81c08ed1594b90b760
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Mon Jan 19 11:39:53 2009 -0600
Add regression test for leading-space messages
There was a long-standing bug in telepathy-idle where a received message that
had a leading space was not parsed correctly and was dropped silently. This bug
has recently been fixed, and this is simply a regression test for it.
commit e2f80415081972431eed901679896d0576d99844
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Fri Jan 16 22:30:28 2009 -0600
Add a regression test for messages delivered in proper order
There was a bug where messages that were queued in the same message timeout
(currently 2s) were not delivered in the order they were received. This bug has
been fixed, but I thought it would be useful to add a regression test for it.
commit 94d765abc6474d7ec5bd2727b26e420f6518a4c8
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Fri Jan 16 22:07:11 2009 -0600
Add a very basic test for joining an IRC channel
I'm not sure whether this test really has a lot of value, but it essentially
documents the current behavior
commit 480d3cea4b59e8191c4116d12b71abfdafb5101c
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Fri Jan 16 17:16:40 2009 -0600
call Disconnect() asynchronously to prevent false test failures
previously, the -success and the -success-ssl tests were failing because they
didn't get a dbus method return from the Disconnect() method. I thought that
this was a problem with idle not returning the response, but it turns out it was
caused by the fact that the test called Disconnect() synchronously. So while
the test was waiting for the dbus return, the QUIT handler could not run, so the
connection was not getting closed. Apparently idle waits for the connection to
be closed before sending the dbus method return.
When I change the Disconnect() call to be async, the test irc client is allowed
to run and the tests now pass.
commit 3fd18659d75157412def057df227e61f4b7aaa6d
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Thu Jan 15 14:23:39 2009 -0600
Ignore a bunch of testing logs
commit 507bf2998e7c8b3cbabf5e1254d24926c07af92c
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Thu Jan 15 11:01:24 2009 -0600
Fix some more minor twisted test issues
- Use expect_many() rather than expect_racy() (the latter was accidentally
committed after I was playing around with it and didn't know exactly what it
did).
- Fix the signature of BaseIRCServer.connectionLost.
- Fix some formatting (tabs vs. spaces) issues
- a little more debug information to help me figure out what's going on.
commit ba450846d8c0ca445ab9fc6263d70c6b2380b651
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Wed Jan 14 12:23:31 2009 -0600
Add support for testing ssl connections with twisted tests
Added a couple ssl connection tests that are just variations of the non-ssl
connection tests and really only serve to excercise the IdleSSLServerConnection
class a bit.
Added a testing-only ssl cert / key just so we're able to start a ssl irc test
server and test connections to it.
commit d9e38485be510aafdd5cb0d870775f6c1d853cd7
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Tue Jan 13 11:18:50 2009 -0600
Improvements to twisted test IRC server
Now at least returns a WELCOME message if it detects a NICK and USER message is
sent by the client. This allows the client to finally transition to the
CONNECTED state from CONNECTING, so it's progress. The test still seems to hang
when attempting to disconnect, though, I'm not sure what is causing that yet.
commit e14ecf6c650b3da01f59094c4d7b74d29543d7e1
Author: Jonathon Jongsma <jjongsma at gnome.org>
Date: Mon Jan 12 14:45:20 2009 -0600
Starting to add twisted test infrastructure
The twisted test infrastructure was copied and modified from gabble. I'm still
getting familiar with it and figuring out how things work, so it doesn't really
do much yet and there are even issues with the few barebones tests I've included
here.
commit 492c8916e6dc4ce7ed82c3509071edd430dada99
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Wed Jan 14 10:15:21 2009 -0600
Fix the fd double-close in the ssl server connection class as well
As in the normal server connection class, there doesn't seem to be any need to
carry the fd around in the AsyncConnectData struct or in the class priv struct
since we can always retrieve it from the GIOChannel when necessary. In
addition, g_io_channel_shutdown closes the fd, so there's no reason to close the
fd separately. In fact, closing the fd separately will result in a glib warning
about closing an invalid file descriptor if the close() is done before the
channel shutdown.
commit d37f1b997bb3dc8a303f6459245072bdcfd3a480
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Tue Jan 13 16:48:16 2009 -0600
Fix a warning about closing an invalid file descriptor
When running the new test suite for idle (not yet merged upstream), I was
getting tests aborted and created core files due to a warning about closing an
invalid file descriptor. The actual source of the warning I was seeing was in
async_connect_data_destroy() while shutting down the IO channel. I believe this
is because a few lines earlier, we close the IOChannel's file descriptor, then
glib attempts to close it again in the g_io_channel_shutdown() function.
We seem to be doing this many other places, but we don't always get warnings
about it because often the close() call comes after the g_io_channel_shutdown()
call, and we don't check the return status from close(). It seems to me that
since glib will free the file descriptor for us when we shut down the IOChannel,
there's not even any reason to carry around the fd separately in
_AsyncConnectData (since we can always get it from the IOChannel anyway), so I
removed that as well.
I haven't changed similar uses in the SSL server connection yet.
commit e798493ed8dc4e5d68035a3362da4e29884d9816
Author: Jonathon Jongsma <jjongsma at gnome.org>
Date: Tue Jan 6 11:47:54 2009 -0600
Handle user renames in muc rooms
Previously, tp_group_mixin_change_members() was called with reason NONE, so UI
applications had no chance of displaying it as a rename rather than separate
quit and join operations. Since idle now depends on a recent-enough tp-glib, we
can use the RENAMED reason (and also, use the new set_self_handle() API in the
TpGroupMixin as well)
commit 022ea3809759d8b1fd92c23b2b8b4e8c1f3b0116
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Tue Jan 13 11:32:17 2009 -0600
Initialize a couple more GErrors to NULL
Missed a couple uninitialized GErrors in the ssl connection class
commit 307977a5ca0db466e52ed5be5b1ebcbf4799296c
Author: Jonathon Jongsma <jonathon.jongsma at collabora.co.uk>
Date: Mon Jan 12 16:32:21 2009 -0600
Initialize all GError* vars to NULL
We were getting some warnings about an error being set over top of an existing
error, because several GError* variables were not initialized properly to NULL.
commit 86ca78e69f077861a6149f5d795ddc0d0aae9503
Author: Jonathon Jongsma <jjongsma at gnome.org>
Date: Wed Dec 31 15:52:47 2008 -0600
Fix parsing error on leading whitespace (bug #17390)
The basic problem is as follows:
There is an incoming message that looks like the following:
":jonner!n=jonner at jalfrezi.collabora.co.uk PRIVMSG #channelname : foo bar"
When this message gets tokenized, we get the following tokens:
- :jonner!n=jonner at jalfrezi.collabora.co.uk
- PRIVMSG
- #channelname
- :
- foo
- bar
When we get to parsing the chat message portion, our current token is simply
":". There is a check in the code that was meant to prevent zero-length strings
from being added to the parsed arguments, but that check assumes that we'll only
be adding the current token to the parsed args. In reality, we want to consume
the entire remaining string (which may contain multiple tokens) and add it to
args. So instead of checking that the the character immediately after the ':'
is non-NULL in the strdup-ed token, we should really check that the character
immediately after the ':' is non-NULL in the original message string.
Bug was introduced in commit 02cf59fb7b7efebbec39422460ec13ae1e3f5c89
commit f65a516ad01362f834f93cff1f947892c2bce46c
Merge: bc09457... 1917b5a...
Author: Simon McVittie <simon.mcvittie at collabora.co.uk>
Date: Thu Sep 25 18:48:40 2008 +0100
Merge branch 'self-handle-changed'
commit bc0945793cf918318538e30d46b281b1de14b4aa
Author: Florian Steinel <Florian.Steinel at googlemail.com>
Date: Thu Sep 25 19:32:40 2008 +0200
fix compile on 64bit arch
commit 1917b5a0406c614f7b3866b7f90ca2cf4dabe83a
Author: Simon McVittie <http://smcv.pseudorandom.co.uk/>
Date: Wed Sep 24 10:56:27 2008 +0100
Require telepathy-glib 0.7.15 (for tp_base_connection_set_self_handle)
commit c1fa7702c628034af76a9c9ebbfe2818f847e069
Merge: bb058c8... 5ae7920...
Author: Simon McVittie <http://smcv.pseudorandom.co.uk/>
Date: Wed Sep 24 10:55:58 2008 +0100
Merge commit 'wjt/self-handle-changed' into self-handle-changed
Conflicts:
configure.ac
commit 5ae7920bb0282fa080cff5d2db6191dcdca35769
Author: Will Thompson <will.thompson at collabora.co.uk>
Date: Tue Sep 9 17:00:21 2008 +0100
Connection: remove unused variable
commit bb058c8892a91e012e002cd2e5526e85bc05e86b
Author: Will Thompson <will.thompson at collabora.co.uk>
Date: Mon Sep 1 12:50:17 2008 +0100
Rewrite idle_text_encode_and_split to be more obviously correct.
Fixes an infinite loop for ACTIONs containing newlines.
commit 06123a3471a6a248604bc18e71221216b1e771e7
Author: Will Thompson <will.thompson at collabora.co.uk>
Date: Mon Sep 1 12:47:13 2008 +0100
Add a test that over-long/newlined messages are split correctly.
commit 8fae440479816222cf1375b89ea1070292ae308b
Author: Will Thompson <will.thompson at collabora.co.uk>
Date: Fri Aug 29 19:26:18 2008 +0100
Send messages with equal priority in chronological order
commit ec8d4ba71c87bab82818bafd4df6bfc5bdacb6f5
Author: Will Thompson <will.thompson at collabora.co.uk>
Date: Fri Aug 29 18:27:18 2008 +0100
Extract splitting of long/multiline messages to a function
commit e7268ab6601c35939e93ab70619b76327a0e1d7f
Author: Will Thompson <will.thompson at collabora.co.uk>
Date: Fri Aug 29 18:04:44 2008 +0100
Fixed signedness warning in a test
commit 7f1faae20c5e8566519fe1e4faa5854a238e3d00
Author: Will Thompson <will.thompson at collabora.co.uk>
Date: Fri Aug 29 17:13:01 2008 +0100
Fix another signedness warning.
commit 49d014ee95dfef2d5d56c2565cdd0c2b2030b190
Author: Will Thompson <will.thompson at collabora.co.uk>
Date: Fri Aug 29 16:59:23 2008 +0100
Correct more mismatched signednesses
commit e86db64ce63408f00d5bd988545984ebac0c3c91
Author: Will Thompson <will.thompson at collabora.co.uk>
Date: Fri Aug 29 16:47:48 2008 +0100
Don't use an invalid enum value as a failure flag
commit 6d2cdaa645d0b626d719aa8edb444eddc602a7e5
Author: Will Thompson <will.thompson at collabora.co.uk>
Date: Fri Aug 29 16:42:28 2008 +0100
Correct the order of "const static" to "static const"
commit c52271ba4cf6425b0cf09f98254d35eaeccac372
Author: Will Thompson <will.thompson at collabora.co.uk>
Date: Fri Aug 29 16:41:52 2008 +0100
Use post-0.6.1 tp_debug_set_flags() and tp_debug_set_persistent()
commit 59381efdfa8e85ce688190b461159873b8f8b708
Author: Will Thompson <will.thompson at collabora.co.uk>
Date: Fri Aug 29 16:41:41 2008 +0100
Bump telepathy-glib dependency to 0.6.1
commit 667dfdbc78652c667513ce581a76d4e6ef5570f0
Author: Will Thompson <will.thompson at collabora.co.uk>
Date: Fri Aug 29 16:08:58 2008 +0100
Bump telepathy-glib dep to 0.7.3 for the DBusPropertiesMixin
commit cefc2b085e80de168abc1e5eef3e867be96dfe5f
Author: Will Thompson <will.thompson at collabora.co.uk>
Date: Fri Aug 29 16:00:28 2008 +0100
Change self-handle using tp_base_connection_set_self_handle
commit 4121b3a3ffcb83f1ecd7797b9e764a9898846c0e
Author: Will Thompson <will.thompson at collabora.co.uk>
Date: Fri Aug 29 16:31:28 2008 +0100
Bump automake dependency
commit 4a25b1496287a2edffc77b99b78e83fbf24f2ed7
Author: Will Thompson <will.thompson at collabora.co.uk>
Date: Fri Aug 29 15:47:46 2008 +0100
Fix a mismatched signedness warning
commit 1bee98f16638f83052ade1213d656b8b87130bed
Author: Will Thompson <will.thompson at collabora.co.uk>
Date: Fri Aug 29 15:47:11 2008 +0100
Enable -Werror and friends on development builds
commit 42d9ae1c93b52f0eba9fee38c739d5e4800a734e
Author: Will Thompson <will.thompson at collabora.co.uk>
Date: Fri Aug 29 15:47:58 2008 +0100
Make autogen.sh executable
commit 9b09b52959ddfc9cab4b29f2434d09656d2516fa
Author: Will Thompson <will.thompson at collabora.co.uk>
Date: Fri Aug 29 15:46:19 2008 +0100
Remove .git-darcs-dir files
commit e2429d549a52c5f7b157688c883b69ac717975e0
Author: Will Thompson <will.thompson at collabora.co.uk>
Date: Fri Aug 29 15:45:54 2008 +0100
Convert _boring to .gitignore
commit 7be13f764bc647755f67e65a24709c36625d16ae
Author: Olli Salli <olli.salli at collabora.co.uk>
Date: Tue Jan 1 19:54:42 2008 +0000
Use g_set_error instead of *error = g_error_new where possible to facilitate
deliberately ignoring errors
20080101195442-9db4d-234ac21aaee90fe01bbcd0c1069086139c67470c.gz
commit fb535cd02fb8baeeb8d0dec46dfe19ce0b87a759
Author: Olli Salli <olli.salli at collabora.co.uk>
Date: Wed Jan 16 16:23:19 2008 +0000
Still more coding style unifications, now there should be quite a few less function
declarations and calls with a space between the function name and the opening paren.
20080116162319-9db4d-8fe36cffc8d4ae36ce744b987a657771b32c98ed.gz
commit 5f954c012af01becad485a8031985c2e2c294582
Author: Olli Salli <olli.salli at collabora.co.uk>
Date: Wed Jan 16 16:05:15 2008 +0000
More coding style unifications. This time make sure all control keywords have a space
between them and the opening paren.
20080116160515-9db4d-ce4b4f35a86d48a12e6c5647eabd93978a5fd081.gz
commit df96afe50ac242e57e5ad2a65ec7839831f8d066
Author: Olli Salli <olli.salli at collabora.co.uk>
Date: Tue Jan 15 21:02:32 2008 +0000
Make GInterface init functions conform to the same coding standard
20080115210232-9db4d-31348a0b3c340fb081133117243a61cbdc995f8f.gz
commit d6884fde268175fca9afccfad0c36435be5a8bd9
Author: Olli Salli <olli.salli at collabora.co.uk>
Date: Tue Jan 15 20:53:29 2008 +0000
Clean up includes
20080115205329-9db4d-6a57ff858a5ce98ae86074e81f3be814e05d6bd3.gz
commit 3c20851336897e76c48a76e6681dc1d64e1c03ac
Author: Simon McVittie <simon.mcvittie at collabora.co.uk>
Date: Thu Jan 3 16:49:55 2008 +0000
Cope with TpChannelIface channel-type property becoming writeable in telepathy-glib >
0.7.0, avoiding warnings when channels are constructed (fd.o #13896).
If telepathy-glib is <= 0.7.0, the added code will never execute, but is
harmless.
20080103164955-53eee-b66b94c229995403a3f178753326941d1bfc623f.gz
commit c077f7ab1f0d49f1d958fc3a8c7bb9b0a191ccc3
Author: Simon McVittie <simon.mcvittie at collabora.co.uk>
Date: Thu Jan 3 16:47:12 2008 +0000
Remove unnecessary reference to an empty variable
20080103164712-53eee-6fe5a4c7cce69a23372eb664a40efce06205ebe0.gz
commit 131fe14ec1f3d6013159739b1e5007bdf96d9d94
Author: Olli Salli <olli.salli at collabora.co.uk>
Date: Sat Jan 12 21:06:59 2008 +0000
Remove useless send_irc_cmd() from idle-connection.c and rename send_irc_cmd_full to
_send_with_priority
20080112210659-9db4d-70b106707c5f33eb0983be516ebf3763ac699523.gz
commit 5cecd8a62742c52fe94838dc30a1d27ef0b41be1
Author: Olli Salli <olli.salli at collabora.co.uk>
Date: Sat Jan 12 20:53:48 2008 +0000
Coding style, part: Public functions of classes shouldn't have an underscore prefix
20080112205348-9db4d-dfcfe6f9e50d5cfda568b3c074d11b62cd44678c.gz
commit aba4d48dc8944a57aadcb5309db8849614c51806
Author: Olli Salli <olli.salli at collabora.co.uk>
Date: Sat Jan 12 20:39:37 2008 +0000
Remove GTK_DOC_CHECK from configure.ac, idle doesn't do gtk-doc
20080112203937-9db4d-a750e2d3b01f65f46119523d67fda7ddc6bda8e7.gz
commit 0134f85f12fe36f8418488e76d78443ceec5ed5e
Author: Olli Salli <olli.salli at collabora.co.uk>
Date: Sat Jan 12 20:38:28 2008 +0000
Add -Wno-portability to autoreconf invocation in autogen.sh, we depend on GNU make
20080112203828-9db4d-19de5ecafe13f70ad99cbec4733b8242d73c6459.gz
commit 8816b1c4afda68b47d799ce82309f393bd4bd5f4
Author: Olli Salli <olli.salli at collabora.co.uk>
Date: Sat Jan 12 12:49:31 2008 +0000
Add HACKING file
20080112124931-9db4d-e97cf76aa09d551c96ef7e543867ddb26aa79340.gz
commit b98a43649d25ef349bb73b8934453f984dc41d9e
Author: Olli Salli <olli.salli at collabora.co.uk>
Date: Thu Dec 27 22:02:08 2007 +0000
Unify coding style, chapter Six: Binary operators and their operands should have
whitespace in between
20071227220208-9db4d-2e018ea7b69ed08e7f17c909b2fa47470c4081e0.gz
commit 64af22ce42221f71c19d2e13ab2df67ee09e5ac4
Author: Olli Salli <olli.salli at collabora.co.uk>
Date: Thu Dec 27 21:36:51 2007 +0000
Fix too greedy removal of cruft resulting in GValue -> GValu
20071227213651-9db4d-ac07090814d0518310802bb5e974451528c35a20.gz
commit 4de1f9c04780edcaecc4247d8fda573b2254d487
Author: Olli Salli <olli.salli at collabora.co.uk>
Date: Thu Dec 27 20:33:00 2007 +0000
Unify coding style, lesson 4: blank lines after each case body in switch statements
20071227203300-9db4d-47a1011afb9c3cd20b58f249b756744115b0b181.gz
commit 60ab6f50523cca46a5d5a03211c26a49dc9173e6
Author: Olli Salli <olli.salli at collabora.co.uk>
Date: Thu Dec 27 20:28:31 2007 +0000
Unify coding style, part 3: Return type on the same line with the function name
20071227202831-9db4d-6fc9278dde6e99383986b366dabbf982c8d4a340.gz
commit af8e39e88f5541edb9f8aa5ab04bc4ef64caad95
Author: Olli Salli <olli.salli at collabora.co.uk>
Date: Thu Dec 27 20:24:21 2007 +0000
Unify coding style, book II: Thou shalt indent by tabs, and only tabs
20071227202421-9db4d-7fb30eae03df9f5ae881719a257c90f3adcb3430.gz
commit 101e2c5b2b5d11491d679e623066dd775b44bcf1
Author: Olli Salli <olli.salli at collabora.co.uk>
Date: Thu Dec 27 20:18:13 2007 +0000
Remove tabs from places other than beginning of lines
20071227201813-9db4d-818203d16e295aa5bd678dd62cccffd76c836f17.gz
commit b1bc692f0ca998950d505c3f4a503eb6899b2258
Author: Olli Salli <olli.salli at collabora.co.uk>
Date: Thu Dec 27 20:14:23 2007 +0000
Remove trailing whitespace
20071227201423-9db4d-9747ce7d1faa02cd80ab78daeec4d2abad6af37f.gz
commit c460df04d67bbfc39cca7e6bbaa4defaf56e76c2
Author: Olli Salli <olli.salli at collabora.co.uk>
Date: Thu Dec 27 20:10:31 2007 +0000
Remove old not-in-use sofia resolver cruft
20071227201031-9db4d-a075bd3de5fed303ce61ab4431e71c0d9a22e09f.gz
commit 0a81e75a123eee5fb4f49d5f9fe7d0590d3571fb
Author: Olli Salli <olli.salli at collabora.co.uk>
Date: Thu Dec 27 20:02:51 2007 +0000
Unify coding style part 1 (mostly bracing style)
20071227200251-9db4d-5f34b917f7042957c8d441c09d09bfe35619d47f.gz
commit 929901419cdf62466ffeec49232a3daf4beb9187
Author: Simon McVittie <simon.mcvittie at collabora.co.uk>
Date: Thu Nov 29 13:14:26 2007 +0000
Stop checking for dbus-binding-tool - no longer needed
20071129131426-53eee-8859a0c37363c61c56ee56a6d2c4577a0c94cbb3.gz
commit 39c65ddf25cf9a8551edb26c84d57528e186f1b6
Author: Simon McVittie <simon.mcvittie at collabora.co.uk>
Date: Thu Nov 29 13:09:08 2007 +0000
Simplify extensions/Makefile.am now that generated GInterfaces don't need external glue
20071129130908-53eee-66c7d1b9d2d05dba1da8e83308d04569d6a7f34b.gz
commit e37b57ae2cc68266d08fc7b7dc5b04317896a17d
Author: Simon McVittie <simon.mcvittie at collabora.co.uk>
Date: Thu Nov 29 13:08:41 2007 +0000
Update tools from telepathy-glib: generated GInterfaces no longer need external glue
20071129130841-53eee-54a96f08127cfd5508f14a03b102ee563638853f.gz
commit d8695095978882927ecf9d58301a1c6ce079f4a5
Author: Simon McVittie <simon.mcvittie at collabora.co.uk>
Date: Thu Nov 29 13:06:10 2007 +0000
Add missing identity.xsl
20071129130610-53eee-6b56bd3209cb7ab359cdfc90ace747eef9b05d1d.gz
commit 9279a0091cdce5b4ca05be6265aeb1120beb6e92
Author: Simon McVittie <simon.mcvittie at collabora.co.uk>
Date: Thu Nov 29 13:02:20 2007 +0000
Add missing glib-signals-marshal-gen.py in tools/
20071129130220-53eee-e98ee6ca716440ffdd4fb9168f87eba964c8de31.gz
commit 06bc54d3d3a0e10e8ea482cba2a145a90be0a8c8
Author: Simon McVittie <simon.mcvittie at collabora.co.uk>
Date: Thu Nov 29 13:01:58 2007 +0000
_boring: ignore test binaries
20071129130158-53eee-b1ddfc54ec9823bbb3a5571162515aaf2492577a.gz
commit a6b79731b78a40d2edf8f689f29e05289783f88f
Author: Simon McVittie <simon.mcvittie at collabora.co.uk>
Date: Fri Nov 23 14:49:43 2007 +0000
Remove _gen/all.xml in clean
20071123144943-53eee-ee4a67796d1eeb688ed0b5720301cf2458745acf.gz
commit 0824ae1319cf55b41c0642ba1874faf85173e85d
Author: Simon McVittie <simon.mcvittie at collabora.co.uk>
Date: Fri Nov 23 14:47:03 2007 +0000
Adjust extensions/Makefile.am: _gen/all.xml should mkdir _gen first
20071123144703-53eee-96fb2aa13569f40abd981b1a80c49caac586a93f.gz
commit fbc95d6db68f565680274efa9fd14eee6000d60c
Author: Simon McVittie <simon.mcvittie at collabora.co.uk>
Date: Fri Nov 23 14:43:55 2007 +0000
Update build system to resemble that of telepathy-glib 0.7
20071123144355-53eee-1a7f7c4dd601a2a9e18288222ecd3dda67b22f41.gz
commit c866efbadc592b9f3d1f883aab2c4bf7d47f1dad
Author: Simon McVittie <simon.mcvittie at collabora.co.uk>
Date: Fri Nov 23 14:40:27 2007 +0000
Check for, and use, mkdir -p or equivalent
20071123144027-53eee-0d7b50c0ec283047b8d391bc8c6c10ca446f1524.gz
commit 4c2af08b47be195a57ac3bb5d9fefe531ea59812
Author: Simon McVittie <simon.mcvittie at collabora.co.uk>
Date: Fri Nov 23 14:17:27 2007 +0000
Sync tools from telepathy-spec
20071123141727-53eee-42fdb5e7b9385290998ffd1c34c8e029dc63dd35.gz
commit ea5b4c00414025ad3de2d4d47f23e8240db70370
Author: Simon McVittie <simon.mcvittie at collabora.co.uk>
Date: Fri Nov 23 14:17:21 2007 +0000
Reference external type Contact_Handle (which is in the Tp spec)
20071123141721-53eee-4e1f305adb423a295a890cc5f7b5d3e2ca729b98.gz
commit 90e3e3d1cc3cf5583841227fef0c0ecf60571958
Author: Simon McVittie <simon.mcvittie at collabora.co.uk>
Date: Fri Nov 23 14:12:41 2007 +0000
Sync tools from telepathy-glib
20071123141241-53eee-1c5b2e4561b000c91190ac8237ff60243c9d266c.gz
commit b2fae91d6a2acf4e7d1ac30cfa61ab529fc15f3d
Author: Simon McVittie <simon.mcvittie at collabora.co.uk>
Date: Fri Nov 23 14:04:02 2007 +0000
Remove update-spec-gen-am.sh
20071123140402-53eee-f7d1a253becb25dc271274bfb389bdea68037e85.gz
commit 031607bd26b6cd7b042ff26eaa0b97d3cf387164
Author: Simon McVittie <simon.mcvittie at collabora.co.uk>
Date: Fri Nov 23 14:01:39 2007 +0000
Move tools directory to top level. Stop using spec-gen.am
20071123140139-53eee-56849a8ae9ca447e90b62de8d14ec6e52f03546c.gz
commit c81f69f6db3db739f2c17ad9c0075e19de3c6f62
Author: Simon McVittie <simon.mcvittie at collabora.co.uk>
Date: Mon Nov 5 11:09:42 2007 +0000
Add telepathy-idle.8 man page
20071105110942-53eee-e69976f4df1c66a993b3c7a1c177540ccdf40e51.gz
commit e5ad1bc2c7c4010d96550f16ea3f7b7ec1687410
Author: Simon McVittie <simon.mcvittie at collabora.co.uk>
Date: Fri Oct 26 11:04:38 2007 +0000
Move CM executable to $libexecdir, adjust .service accordingly
20071026110438-53eee-867b50d1ec1e62cba39f29b9208beb5e470bde53.gz
commit c2d95a4b96eae6f5e692ace9f200589695cc6f6f
Author: Simon McVittie <simon.mcvittie at collabora.co.uk>
Date: Fri Sep 28 13:27:46 2007 +0000
Set nano version 0.1.2.1
20070928132746-53eee-2d8a25cddd9dda524317cc6cf395af50e9818796.gz
--
Jonathon Jongsma
More information about the telepathy
mailing list