2.1.x#28
Conversation
|
Thanks, David ! We are reviewing the patch. Most of the time, it's not doing anything unless you are using cached buffers. I assume this is teh case for you, then the buffer aren't getting cached if they aren't freed. |
|
Any idea when this pull request might be accepted into the mina code base?
David R Robison
Senior Systems Engineer
[cid:image001.png@01D6BD8D.119EB360]
From: Emmanuel Lécharny <notifications@github.com>
Sent: Thursday, September 24, 2020 10:08 AM
To: apache/mina <mina@noreply.github.com>
Cc: David Robison <david.robison@psgglobal.net>; Author <author@noreply.github.com>
Subject: Re: [apache/mina] 2.1.x (#28)
Thanks, David !
We are reviewing the patch. Most of the time, it's not doing anything unless you are using cached buffers. I assume this is teh case for you, then the buffer aren't getting cached if they aren't freed.
—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub<#28 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AAYWJXCYW66KNNT5DWVEA3DSHNHENANCNFSM4RYK2BLA>.
|
|
Patch finally applied.... Sorry for the 4 years delay, lots of day job/family things got in the path... |
|
There's a regression in the patch that was pushed separately by @elecharny. 88cb553#diff-450772ac8848cc0e6cc29eef8dec7d3d2b5633a3f6f2c9f676169b3f73f28f6aR242 This changes removeSessionBuffer() to use getAttribute instead of removeAttribute. After this call, the session will still have a reference to the buffer, and if it's a SimpleBuffer, it will not have been freed. This results in doDecode() re-decoding portions of the buffer. The patch from @kc7bfi is correct but was not merged. a8385ae#diff-450772ac8848cc0e6cc29eef8dec7d3d2b5633a3f6f2c9f676169b3f73f28f6aR240 |
|
Checking the {{CumulativeProtocolDecoder}} again, the code is under optimal... Here what I suggest: |
There were some places where buffers were not properly duplicated and/or freed. This caused poor performance when attempting to cache IOBuffers.