Emiel Bruijntjes
|
f23dc72a4f
|
Fixed documentation, removed references to the ChannelHandler - an object that is no longer supported by AMQP-CPP
|
2015-01-26 14:47:30 +01:00 |
|
Emiel Bruijntjes
|
236dd02b13
|
fixed theoretical issue when deferredGet onSize() methods destructs the underlying channel object
|
2014-09-02 11:18:11 +02:00 |
|
Emiel Bruijntjes
|
9653578d4a
|
The Channel.get().onFinalize() function was called before the Channel.get().onSuccess() method, which is strange. This has been fixed. To fix this, we had to refactor the Deferred* into shared_ptrs
|
2014-09-02 10:32:55 +02:00 |
|
Emiel Bruijntjes
|
aceac8b677
|
the finalize callback is now called as part of the deferred object destructor, this ensures that the finalize callback will always be called
|
2014-08-29 13:18:08 +02:00 |
|
Emiel Bruijntjes
|
8759258a3e
|
Channel::get() and Channel::purgeQueue() methods always reported an error, even while there was no error
|
2014-08-29 13:16:05 +02:00 |
|
Emiel Bruijntjes
|
ebcdd71848
|
Added onSize() method to DeferredGet
|
2014-08-28 10:02:01 +02:00 |
|
Emiel Bruijntjes
|
c7b3f71f14
|
the connection.close function was only waiting for synchronous calls to complete, async calls that were waiting (after a synchronous) were still discarded, this has been fixed
|
2014-08-20 13:40:29 +02:00 |
|
Emiel Bruijntjes
|
d23e818f64
|
turns out that shared_from_this() can not be called from the constructor, so we made a special attach() function to postpone object initialization
|
2014-08-20 12:44:52 +02:00 |
Michael van der Werve
|
53b2bd9f0d
|
fixed typo in the connection implementation file
|
2014-08-20 12:03:15 +02:00 |
|
Emiel Bruijntjes
|
36734671cc
|
channel instructions that were given before the connection was fully initialized were immediately lost, if the connection was immediately closed
|
2014-08-20 11:59:05 +02:00 |
|
Emiel Bruijntjes
|
6997a70cf1
|
when a connection was closed, the commands already given to the channels were sometimes by-passed by the closing channel. This has been fixed. Also solved the problem that calls executed on channel objects that fall out of scope were not always completed
|
2014-08-20 11:47:16 +02:00 |
|
Emiel Bruijntjes
|
328820f898
|
renamed apple.h to endian.h because it turns out that it is also used for linux systems, and removed double implemented ChannelImpl::get() function
|
2014-08-14 15:25:44 +02:00 |
|
Emiel Bruijntjes
|
3e47191f90
|
fixed apple incompatibility
|
2014-08-14 15:19:08 +02:00 |
|
Emiel Bruijntjes
|
271cc39c17
|
whitespace change
|
2014-08-13 16:43:59 +02:00 |
|
Emiel Bruijntjes
|
5bc7b4b5b3
|
added destructor of ReducedBuffer
|
2014-08-13 15:02:54 +02:00 |
|
Emiel Bruijntjes
|
945a01f659
|
The Connection::parse() method can now also be called with a buffer object, which allows the user to implement the buffer much more efficiently: it no longer has to be an array
|
2014-08-13 13:01:27 +02:00 |
Richard Hodges
|
d2a84b5f8d
|
build - added new source files to CMakeLists
|
2014-08-07 11:09:02 +01:00 |
|
Emiel Bruijntjes
|
a8ff6de550
|
fixed bug in channel.get() calls
|
2014-08-07 11:00:06 +01:00 |
|
Emiel Bruijntjes
|
37a51cdc7b
|
basic.get is asynchronous until the entire body has been received, only then subsequent messages are sent + first work in progress on implementing smarter buffers
|
2014-08-07 11:00:06 +01:00 |
|
Emiel Bruijntjes
|
dca76db2f0
|
empty messages could not be consumed or get, this has now been fixed
|
2014-08-07 11:00:06 +01:00 |
|
Emiel Bruijntjes
|
bcbe50e22c
|
basic.get callbacks only have to be installed for a single message, so we uninstall it right after the message was handled
|
2014-08-07 11:00:06 +01:00 |
|
Emiel Bruijntjes
|
b4270f39bc
|
added Channel::get() that allows one to retrieve a single message from the channel
|
2014-08-07 11:00:06 +01:00 |
|
Emiel Bruijntjes
|
f905c9db49
|
Quality of service now supports "global" parameter (default is still false)
|
2014-08-07 11:00:05 +01:00 |
|
Emiel Bruijntjes
|
d99b117a60
|
fixed bug in channel.get() calls
|
2014-08-04 13:44:17 +02:00 |
|
Emiel Bruijntjes
|
e299aa5fac
|
basic.get is asynchronous until the entire body has been received, only then subsequent messages are sent + first work in progress on implementing smarter buffers
|
2014-08-01 11:55:07 +02:00 |
|
Emiel Bruijntjes
|
b97222c4db
|
empty messages could not be consumed or get, this has now been fixed
|
2014-08-01 10:37:18 +02:00 |
|
Emiel Bruijntjes
|
7956d706fb
|
basic.get callbacks only have to be installed for a single message, so we uninstall it right after the message was handled
|
2014-08-01 10:05:02 +02:00 |
|
Emiel Bruijntjes
|
42514e80db
|
Merge branch 'master' of https://github.com/CopernicaMarketingSoftware/AMQP-CPP
|
2014-07-31 13:11:30 +02:00 |
|
Emiel Bruijntjes
|
f39410039b
|
added Channel::get() that allows one to retrieve a single message from the channel
|
2014-07-31 12:58:13 +02:00 |
Martijn Otto
|
f2f241665f
|
Merge branch 'master' of github.com:CopernicaMarketingSoftware/AMQP-CPP
|
2014-07-31 10:17:15 +02:00 |
|
Martijn Otto
|
98aba8f638
|
Fix for issue #12
|
2014-07-31 10:16:55 +02:00 |
|
Emiel Bruijntjes
|
cac3247304
|
Quality of service now supports "global" parameter (default is still false)
|
2014-07-31 10:10:15 +02:00 |
|
Richard Hodges
|
d4b10cbf35
|
bug - negate the result of send() before using the result in a DeferredResult constructor
|
2014-07-30 15:29:05 +01:00 |
|
Richard Hodges
|
d2a97c1dd3
|
bugfix - fixes bug where channel declaration deferred object is put in fail state after successful send of data
refs #12
|
2014-07-30 09:39:20 +01:00 |
|
Richard Hodges
|
b9ca150dba
|
merged from latest upstream
|
2014-07-28 15:16:32 +01:00 |
|
Richard Hodges
|
b41e2edc52
|
Merge remote-tracking branch 'upstream/master'
|
2014-07-28 15:07:06 +01:00 |
|
Emiel Bruijntjes
|
850252e1dd
|
link time optimization only for static libs
|
2014-07-25 09:34:01 +02:00 |
|
Richard Hodges
|
7c0642f30d
|
build - added CMakeLists.txt to allow integration into a CMAKE build tree
|
2014-07-11 18:17:03 +01:00 |
|
Emiel Bruijntjes
|
3a700226c8
|
BasicNack frame makes the consumer stop (dont know why) so we use the BackReject message instead when there is no need to reject multiple messages
|
2014-05-26 18:17:49 +02:00 |
|
Martijn Otto
|
a9570277b7
|
Removed the nowait option from the public interface, because the deferred would never be called and implemented a queue to wait for synchronous methods to complete before sending the next frame
|
2014-04-29 15:51:33 +02:00 |
|
Emiel Bruijntjes
|
1fecc57d67
|
when a connection gets in error state, all deferred results will now also call their error callback, and the channel wide error handler is called _after_ all individual error handlers are called
|
2014-04-16 09:25:08 +02:00 |
|
Emiel Bruijntjes
|
82249ee368
|
update documentation, error callbacks now get a const char * instead of a std::string
|
2014-04-15 14:22:30 +02:00 |
|
Emiel Bruijntjes
|
3d4a1b865e
|
removed some @todo tags
|
2014-04-15 13:22:06 +02:00 |
|
Emiel Bruijntjes
|
b13398b09d
|
setTimeout function removed from connection handler, the finalize and error callbacks are called right away if installed on an object that already is in an error state
|
2014-04-15 13:18:32 +02:00 |
|
Emiel Bruijntjes
|
60b59524e7
|
when an error is detected on a channel, all subsequent and cached deferred objects are notified about the error too
|
2014-04-15 13:14:16 +02:00 |
|
Emiel Bruijntjes
|
d08270701e
|
refactored dealing with error messages
|
2014-04-15 13:01:27 +02:00 |
|
Emiel Bruijntjes
|
ae7a32a8bf
|
when a consumer is cancelled, it is also removed from the map of active consumers in the the ChannelImpl object
|
2014-04-15 12:36:11 +02:00 |
|
Emiel Bruijntjes
|
3b78247363
|
error callbacks get a const char *, no longer a std::string
|
2014-04-15 12:29:22 +02:00 |
|
Emiel Bruijntjes
|
301b8153e3
|
deferred objects are now correctly destructed + added extra checks so that no crashes occur when someone destructs a channel inside a callback function
|
2014-04-15 12:25:56 +02:00 |
|
Emiel Bruijntjes
|
745ab512a5
|
the consumer message callback can now also be installed via the Deferred objects, and it is no longer passed a consumer tag, because it already is obvious what the consumer tag is supposed to be
|
2014-04-15 11:39:52 +02:00 |