Trustin Lee
|
8d5d8fd172
|
Fixed another race condition which occurs when OP_WRITE and OP_READ flags are set/cleared at the same time
|
2008-10-01 08:42:26 +00:00 |
|
Trustin Lee
|
cfa6794292
|
Potential fix for NETTY-52 (IllegalArgumentException when trying to change the current interestOps)
|
2008-10-01 05:25:01 +00:00 |
|
Trustin Lee
|
731ab2a0eb
|
Updated copyright information
|
2008-10-01 04:55:10 +00:00 |
|
Trustin Lee
|
dae3b05ebb
|
Replaced FastQueue with LinkedTransferQueue
|
2008-10-01 04:49:10 +00:00 |
|
Trustin Lee
|
6d372c367b
|
Added LinkedTransferQueue
|
2008-10-01 03:18:36 +00:00 |
|
Trustin Lee
|
2f4e20ce7a
|
Made more generic
|
2008-09-30 14:23:38 +00:00 |
|
Trustin Lee
|
44c409ca24
|
More accurate default receive buffer size predictor implementation
|
2008-09-30 13:02:41 +00:00 |
|
Trustin Lee
|
0831b32090
|
Decreased the default maximum receive buffer size to 64KiB - it's enough for more cases.
|
2008-09-30 06:29:16 +00:00 |
|
Trustin Lee
|
950a386d3f
|
Code cleanup
|
2008-09-30 01:53:24 +00:00 |
|
Trustin Lee
|
7de92967c2
|
Made MemoryAwareThreadPoolExecutor.shouldCount() extensible
|
2008-09-30 01:52:02 +00:00 |
|
Trustin Lee
|
85dc8f93f5
|
Fixed a bug where per-channel memory limitation didn't work correctly
|
2008-09-30 01:49:22 +00:00 |
|
Trustin Lee
|
8c15102bc3
|
Fixed issue NETTY-51 (MemoryAwareThreadPoolExecutor and its subtypes should use long instead of int to store the counters)
|
2008-09-30 00:42:04 +00:00 |
|
Trustin Lee
|
6f18c940dd
|
Fixed issue NETTY-50 (Dead lock in MemoryAwareThreadPoolExecutor)
* Made sure that the max memory size properties become immutable
|
2008-09-30 00:31:32 +00:00 |
|
Trustin Lee
|
50f043fa3e
|
Fixed issue NETTY-50 (Dead lock in MemoryAwareThreadPoolExecutor)
* Fixed incorrect release condition
|
2008-09-30 00:16:42 +00:00 |
|
Trustin Lee
|
d7c53437e9
|
Fixed errors in JavaDoc about OIO-based transport shutdown
|
2008-09-29 06:51:41 +00:00 |
|
Trustin Lee
|
8fb2f7c789
|
It's actually enough with a simple shutdown lock because we have only one acceptor thread.
|
2008-09-28 15:07:47 +00:00 |
|
Trustin Lee
|
dccc9f8665
|
* Improved the throughput of the server-side accept operation
* Added FastQueue.isEmpty()
|
2008-09-28 15:01:21 +00:00 |
|
Trustin Lee
|
4d17db6eb1
|
Using LinkedList seems to yield same or better performance
|
2008-09-28 14:21:38 +00:00 |
|
Trustin Lee
|
40b1f26c34
|
Made sure all segments are evaluated on poll for potentially better throughput (still proof-of-concept - getting better)
|
2008-09-28 14:12:28 +00:00 |
|
Trustin Lee
|
597cbd870c
|
Isolated unchecked cast warning from FastQueue.poll()
|
2008-09-28 13:57:17 +00:00 |
|
Trustin Lee
|
6611563c8c
|
Fixed a race condition in FastQueue
|
2008-09-28 13:52:49 +00:00 |
|
Trustin Lee
|
5170838e79
|
Added direct buffer support (disabled by default and can't be enabled without recompilation for now - should be fixed in 3.1.)
|
2008-09-28 12:51:50 +00:00 |
|
Trustin Lee
|
b64124efd6
|
* Renamed WriteMessageQueue to FastQueue, made it generic and optimized it
* Added ConcurrentFastQueue (proof of concept yet - slow)
|
2008-09-27 14:22:52 +00:00 |
|
Trustin Lee
|
a0d132a4d8
|
Moved WriteMessageQueue to the util package
|
2008-09-27 11:53:00 +00:00 |
|
Trustin Lee
|
f2c1734636
|
Reduced contention between I/O thread and writer
|
2008-09-26 12:28:09 +00:00 |
|
Trustin Lee
|
90ea9aa047
|
Formatting
|
2008-09-26 09:05:41 +00:00 |
|
Trustin Lee
|
362ecf640b
|
Fixed a bug where DiscardClient didn't work since the last optimization
|
2008-09-26 09:04:02 +00:00 |
|
Trustin Lee
|
39086edae6
|
Introduced an intermediary write buffer to reduce the contention
|
2008-09-26 08:31:33 +00:00 |
|
Trustin Lee
|
0241120ace
|
Optimization.. (reduced volatiles)
|
2008-09-26 07:00:30 +00:00 |
|
Trustin Lee
|
22d79a501d
|
Replaced ConcurrentLinkedQueue with LinkedBlockingQueue in NioSocketChannel
|
2008-09-26 04:31:41 +00:00 |
|
Trustin Lee
|
9f5468f7a7
|
Made sure NioWorker.taskQueue doesn't contain the same WriteTask
|
2008-09-26 03:32:26 +00:00 |
|
Trustin Lee
|
5f1ecc9022
|
Fixed throughput issue when messages are written from different threads than I/O worker threads
|
2008-09-26 03:02:31 +00:00 |
|
Trustin Lee
|
ebdcb21421
|
Made sure StackTraceSimplifier is always loaded
|
2008-09-26 02:57:10 +00:00 |
|
Trustin Lee
|
824e0034c4
|
A little bit of optimization
|
2008-09-26 00:43:14 +00:00 |
|
Trustin Lee
|
e1f4053d41
|
At last but not least
|
2008-09-25 16:31:03 +00:00 |
|
Trustin Lee
|
33c1932353
|
Dunno why but SocketChannel.accept() sometimes returns null.
|
2008-09-25 11:05:45 +00:00 |
|
Trustin Lee
|
bc60f01b58
|
section -> chapter
|
2008-09-24 12:38:52 +00:00 |
|
Trustin Lee
|
9c0b5d1949
|
Better table L&F
|
2008-09-24 12:01:50 +00:00 |
|
Trustin Lee
|
87390319c8
|
Added the list of available options
|
2008-09-24 11:28:46 +00:00 |
|
Trustin Lee
|
274fa4aac5
|
A little bit of duplication to help understanding of asynchronous operation
|
2008-09-24 10:49:39 +00:00 |
|
Trustin Lee
|
d0d99ff6bc
|
More detailed explanation on ChannelFuture
|
2008-09-24 10:46:58 +00:00 |
|
Trustin Lee
|
320b7133a2
|
More clarification on ChannelHandlerContext
|
2008-09-24 10:37:19 +00:00 |
|
Trustin Lee
|
92be78df49
|
Rephrased the description of ChannelHandlerContext
|
2008-09-24 10:08:30 +00:00 |
|
Trustin Lee
|
60eec52dab
|
Typo
|
2008-09-24 09:56:10 +00:00 |
|
Trustin Lee
|
0b8687cdeb
|
More explanation on the characteristics of bootstraps
|
2008-09-24 09:55:19 +00:00 |
|
Trustin Lee
|
008f7c394e
|
Added the PNG version of the architecture diagram
|
2008-09-24 09:50:38 +00:00 |
|
Trustin Lee
|
aaa6efcda4
|
Fires -> Sends
|
2008-09-24 09:49:23 +00:00 |
|
Trustin Lee
|
6a759f0c63
|
Fixed grammar, typo, and brevity in Javadoc comments
|
2008-09-24 09:48:32 +00:00 |
|
Trustin Lee
|
ec06cbf57a
|
Made the diagram more compact
|
2008-09-24 05:32:43 +00:00 |
|
Trustin Lee
|
880a093c9f
|
Added the architecture diagram
|
2008-09-24 04:45:39 +00:00 |
|