d76b0a9aa2
Merged default into bug21172
2009-09-29 05:29:17 +01:00
aebd4c40b3
Merged bug 20399 into default
2009-09-28 17:25:19 +01:00
65202e832f
merging from default
2009-09-28 17:16:33 +01:00
3304079373
an internal error in the channel process now takes the entire connection down
2009-09-28 14:48:19 +01:00
fc2e371d6f
merging from default
2009-09-25 13:46:28 +01:00
5058e2f225
Merged bug21618 into default
2009-09-25 10:07:03 +01:00
f51a6bd49e
refactoring handling of EXIT signals in amqp_connection; other cosmetic
2009-09-24 21:50:18 +01:00
a835028768
fixing race condition between writer0 and main_reader exit signals; other small fixes
2009-09-24 21:07:49 +01:00
21bf3ce6ae
reader doesn't need trap_exit
2009-09-23 19:42:37 +01:00
d6e223b981
drivers must not be aware of channel_state (forgot about close_channel)
2009-09-23 19:03:22 +01:00
ace45bfa87
drivers must not be aware of channel_state
2009-09-23 17:00:58 +01:00
6feeba38a3
taking care of renamed field in #amqp_error{}; making naming of variables from amqp_error{} fields consistent
2009-09-22 13:51:22 +01:00
a217ab14c9
changing error reasons of type {amqp, ...} to #amqp_error{}
2009-09-17 14:21:20 +01:00
459c294a43
removing unnecessary relinking of writer process
2009-09-11 18:37:15 +01:00
6d52741740
closing rabbit framing channels properly
2009-09-11 18:32:24 +01:00
35356e4143
framing channel is now started from the channel process and monitored by the main reader; refactoring exit trapping in amqp_connection; refactoring channel registration in amqp_connection
2009-09-11 17:15:38 +01:00
e9ebd15746
merging from default
2009-09-11 13:44:07 +01:00
6358042351
refactoring open_channel: now mostly done in the channel process - simpler, more readable; refactoring of managing the framing channel to follow
2009-09-11 13:41:03 +01:00
0d26dde4bd
Merged bug21616 into default
2009-09-10 23:13:31 +01:00
e3184ce399
refactoring framing channel registration in the reader process; handling socket_closed exit in amqp_channel separately; linking framing channel with reader
2009-09-10 19:41:33 +01:00
191fc90cea
merging from default
2009-09-10 13:35:04 +01:00
75e7f1fe5c
Merged bug21563 into default
2009-09-09 10:41:50 +01:00
1f06924fb8
refactoring amqp_channel:rpc_top_half and rpc_bottom_half
2009-09-08 21:01:39 +01:00
10a5a97a86
merging bug21533 into default
2009-09-07 12:33:50 +01:00
1d46406bac
fixing merge conflict
2009-09-06 20:35:07 +01:00
c322a9fa94
updating from default
2009-09-06 20:32:35 +01:00
51a9bb133e
merging bug21522 into default
2009-09-06 20:01:59 +01:00
89b3239aab
Qualified the exit from the channel0 process
2009-09-03 18:22:52 +01:00
2500df69e9
Refactored the channel registration in the connection process so that it reads better and allocates channel numbers in a more sensible fashion. channel processes that die on the client are trapped and the connection process just ignores them, i.e. the channel number is not de-allocated, because that would free up the channel number and hence confuse the server. This bug requires 21522 and 21533 to land before this can be seriously tested.
2009-09-02 01:32:24 +01:00
b3fa3fb431
Added shutdown statements to tests that were leaking processes
2009-09-01 16:21:31 +01:00
684312d70e
Introduced a closing state when the broker initiates the shutdown. This allows
...
the socket reader process to wait for the broker to close the socket (or
timeout) before stopping amqp_connection process. Squirrel the close
reason away in the state of the connection process. The resulting EXIT message
from the reader process is trapped and then the close reason that was
squirreled away can is used as the EXIT reason of the amqp_connection
process.
Because the amqp_channel process links to the connection process, it will
receive the same EXIT reason and return it to whatever application
process is using the amqp_channel process.
The network driver sees some minor defintion cleanups as well as removing
the link between the amqp_channel_process and the rabbit_framing_channel
process -
this is because the amqp_channel is not able to sensibly handle any exit
the framing channel will send.
Made the teardown routine assert that the channel and connection
processes are in actual fact dead after having closed the connection for every test.
The connection process does not shutdown synchronously.
Also put some repetitions into the tests that were experiencing race conditions.
2009-09-01 13:47:40 +01:00
6a7267bc5f
erhm. > 5.7.0 is good enough
2009-08-29 15:40:26 +01:00
e012b9a785
make type spec Erlang version constraint match that of the server Makefile
2009-08-29 15:32:19 +01:00
9fcd022865
Merged bug21404 into default
2009-08-28 15:02:24 +01:00
0b1bda4bf2
Merged default into bug21404
2009-08-28 13:33:57 +01:00
ae864fc214
Minor tweak to get the app descriptor to get created
2009-08-28 11:59:55 +01:00
1a6327f8b3
Merged bug21517 into default
2009-08-28 11:13:36 +01:00
dd39c007ec
Merged bug21488 into default
2009-08-28 11:05:34 +01:00
7639f5e7c8
Merged bug21408 into default
2009-08-28 10:20:41 +01:00
b73ca87503
Updated application descriptor to use RabbitMQ AMQP Client as the name
2009-08-28 10:14:14 +01:00
a3fcb19431
Merged bug21515 into default
2009-08-28 09:00:56 +01:00
0ae77575f9
Fixed missing copy of the .hrl file
2009-08-28 08:59:37 +01:00
a88750f220
Added .app files to the ignore list
2009-08-28 08:52:35 +01:00
2f5010d11a
Call to channel.close now sets the closing flag
2009-08-27 19:28:47 +01:00
a0407b3670
.app files now are generated by the Makefile
2009-08-27 18:23:17 +01:00
7f27d30fc4
Cherry picked old patch, now uses the app descriptor to get the version
2009-08-27 17:49:20 +01:00
e2cac3825a
Merged default back into bug21488
2009-08-27 16:08:07 +01:00
c7d1554a90
Somehow some io:formats got merged in from the SSL branch I think
2009-08-27 16:05:38 +01:00
ec5fb387e0
Merged default into bug21488
2009-08-27 12:34:48 +01:00
c6b9f23db7
Connection level error in the direct case now is propagated to the channel process
2009-08-27 12:31:54 +01:00
Ben Hood
Marek Majkowski
Vlad Ionescu
Matthias Radestock
Paul Jones