Summary: | NewChannels fired twice when connecting to a bip account | ||
---|---|---|---|
Product: | Telepathy | Reporter: | Guillaume Desmottes <guillaume.desmottes> |
Component: | idle | Assignee: | Telepathy bugs list <telepathy-bugs> |
Status: | RESOLVED FIXED | QA Contact: | Telepathy bugs list <telepathy-bugs> |
Severity: | normal | ||
Priority: | medium | CC: | rishi.is |
Version: | unspecified | ||
Hardware: | Other | ||
OS: | All | ||
Whiteboard: | |||
i915 platform: | i915 features: | ||
Attachments: |
Do not emit an extra NewChannels while handling a JOIN message
0002 Test to cover MUC JOIN initiated by the server 0003 Strengthen the MUC JOIN test to forbid duplicate NewChannels Handle PART messages in our test IRC server 0002 Test to cover MUC JOIN initiated by the server |
Description
Guillaume Desmottes
2010-07-05 03:55:33 UTC
tp_channel_manager_emit_new_channel is fired twice: - In _join_handler() when we receive the JOIN command from bip - In _channel_join_ready_cb() when we actually joined the room I guess we shouldn't emit the first NewChannels signal but I don't know enough of Idle/IRC to be sure that's the proprer fix. The same thing happens when you get invited to a channel that you are not already a part of. Basically whenever a JOIN message is received for a channel that you did not attempt to join yourself. (In reply to comment #2) > Basically whenever a JOIN message is received for a channel that you did not > attempt to join yourself. Oops, sorry! What was I thinking? Here is the correct analysis. Whenever we try to handle an INVITE or a JOIN for which there is no channel, we do (as Guillaume wrote): + _muc_manager_new_channel + tp_channel_manager_emit_new_channel (emits NewChannels) After that, in case of JOIN, irrespective of whether there was a channel or not, we do: + idle_muc_channel_join (emits NewChannels) And in case of INVITE, we try a normal join operation if the user wants to, which leads to a NewChannels as well. I am not sure about the behaviour in case of an INVITE, but for the _join_handler, the easy way out is to not call tp_channel_manager_emit_new_channel. However, I am told that older versions of Bitlbee (http://www.bitlbee.org/main.php/news.r.html) does not want the client to leave the &bitlbee channel and sends a JOIN to the client whenever it does so. I don't know if this is a correct use of JOIN and since newer Bitlbee versions don't do so, I am not sure if we should support this. Created attachment 45793 [details] [review] Do not emit an extra NewChannels while handling a JOIN message Created attachment 45949 [details] [review] 0002 Test to cover MUC JOIN initiated by the server Created attachment 45950 [details] [review] 0003 Strengthen the MUC JOIN test to forbid duplicate NewChannels Although not directly related to this, it might be a good idea to improve the existing channels/join-muc-channel.py to forbid duplicate NewChannels. Review of attachment 45950 [details] [review]: Looks good. Review of attachment 45949 [details] [review]: Could the test also check that PARTing a room, and having the server echo the PART followed immediately by JOINing you again, works? ::: tests/twisted/channels/join-muc-channel-bouncer.py @@ +3,3 @@ +Test connecting to a IRC channel +""" + I think this comment needs updating to match the test! :) Review of attachment 45793 [details] [review]: If all the tests pass, including your new one, then this looks fine! Created attachment 46264 [details] [review] Handle PART messages in our test IRC server (In reply to comment #8) > Review of attachment 45949 [details] [review]: > > Could the test also check that PARTing a room, and having the server echo the > PART followed immediately by JOINing you again, works? Done. > ::: tests/twisted/channels/join-muc-channel-bouncer.py > @@ +3,3 @@ > +Test connecting to a IRC channel > +""" > + > > I think this comment needs updating to match the test! :) Done. Created attachment 46265 [details] [review] 0002 Test to cover MUC JOIN initiated by the server Committed and pushed. |
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.