[
https://issues.apache.org/jira/browse/PROTON-2140?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16983490#comment-16983490
]
Robbie Gemmell commented on PROTON-2140:
----------------------------------------
RE: https://gitbox.apache.org/repos/asf?p=qpid-proton.git;h=6be92fc
The error behaviour clarification text for links is in the wrong place, it was
added on the related condition method, i.e the method folks are actually meant
to use.
Those clarification updates don't fit in well with the preceding text for the
deprecated methods, which first explain how the method works (probably wrong
from some time?) and the how the pointer is valid until the endpoint is freed.
They now follow that essentially by explaining none of what came before is the
case, since it always returns an empty error.
The old methods returned NULL if no link/session/connection was given...even
though it seems unlikely the methods would be called that way, should the new
ones act the same in that regard given they are only there to preserve ABI?
It looks like deprecation isnt noted on the binding APIs at all, and so neither
will the new update be. Worth changing?
More curiosity/aside...
Though proton isnt using the error (except bindings exposing it per above),
what actually happens if other code was [ab]using the per-endpoint error for
its own purposes previously, given the new shared 'empty error'? Segfault?
The deprecation warning for the methods are still commented out. I guess that
is as enabling them would affect compatibility? If so, will it ever actually be
uncommented, or will the method just removed instead at the same point it could
be uncommented?
> proton-c has very high memory footprint for links
> -------------------------------------------------
>
> Key: PROTON-2140
> URL: https://issues.apache.org/jira/browse/PROTON-2140
> Project: Qpid Proton
> Issue Type: Improvement
> Reporter: Gordon Sim
> Priority: Major
> Fix For: proton-c-0.30.0
>
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
