[
https://issues.apache.org/jira/browse/PROTON-1948?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16637118#comment-16637118
]
ASF subversion and git services commented on PROTON-1948:
---------------------------------------------------------
Commit 1a5e1936944c11bae64244a3223e3be98b2d6d18 in qpid-proton-j's branch
refs/heads/master from [~tabish121]
[ https://git-wip-us.apache.org/repos/asf?p=qpid-proton-j.git;h=1a5e193 ]
PROTON-1948 Refactor FrameWriter with more performant buffer
Refactor the FrameWriter to use a more performant ReadableBuffer wrapper
around a byte array that auto grows as an ecode requires instead of
reallocating a ByteBuffer and copying past contents and then performing
a new encode any time space is exhausted. Removing the layer of buffer
abstractions leads to an increase in overall performance under normal
operations as well.
> Refactor FrameWriter to avoid reencodes when buffer space is insufficient
> -------------------------------------------------------------------------
>
> Key: PROTON-1948
> URL: https://issues.apache.org/jira/browse/PROTON-1948
> Project: Qpid Proton
> Issue Type: Improvement
> Components: proton-j
> Affects Versions: proton-j-0.29.0
> Reporter: Timothy Bish
> Assignee: Timothy Bish
> Priority: Minor
> Fix For: proton-j-0.30.0
>
>
> The current FrameWriter implementation uses a ByteBuffer wrapped in a
> ReadableBuffer to encode to and if the buffer is to small it recreates the
> buffer with a larger size and must perform a second encode to finish the job.
> We can now implement our own ReadableBuffer type for the writer that both
> performs better by dropping the ByteBuffer abstractions and that grows as
> needed to fit the encoding of the types and the accompanying payloads (for
> Transfers).
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
