[
https://issues.apache.org/jira/browse/PROTON-2958?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18106698#comment-18106698
]
ASF subversion and git services commented on PROTON-2958:
---------------------------------------------------------
Commit 55c98c724d1f5e9141dfc2e7b5e121683449f526 in qpid-proton's branch
refs/heads/main from Andrew Stitcher
[ https://gitbox.apache.org/repos/asf?p=qpid-proton.git;h=55c98c724 ]
PROTON-2958: Implement small encoding in pn_data_t for lists/maps/arrays
We do this by shifting back the content if we detect it is small enough
at the point where we would go and back fill the size.
A major limitation of this approach is that arrays have to contain identical
typecodes so we assume that everything in one cannot be shortened to avoid a
mix of typecodes.
> Implement small AMQP encodings for list/maps/arrays in pn_data_t encoder
> ------------------------------------------------------------------------
>
> Key: PROTON-2958
> URL: https://issues.apache.org/jira/browse/PROTON-2958
> Project: Qpid Proton
> Issue Type: Improvement
> Components: proton-c
> Affects Versions: proton-c-0.40.0
> Reporter: Andrew Stitcher
> Assignee: Andrew Stitcher
> Priority: Major
> Fix For: proton-c-0.41.0
>
>
> For very small compound structures using the small encodings can save a large
> percentage of on the wire bytes (6 bytes), and for larger ones it's not
> insignificant.
> The simplest implementation would be to detect the small encoding is allowed
> at the point we go back and fill in the size. Then shift the content bytes
> back 6 bytes if allowed before filling in the size and maybe fixing up the
> count. Since the maximum number of bytes to shift is 255 the cost would not
> be too high.
> A major limitation of this approach is that arrays have to contain identical
> typecodes so we assume that everything in one cannot be shortened to avoid a
> mix of typecodes.
>
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
