You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@qpid.apache.org by Gordon Sim <gs...@redhat.com> on 2015/04/17 13:24:44 UTC
Review Request 33300: PROTON-850: allow reusing link names
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/33300/
-----------------------------------------------------------
Review request for qpid and Rafael Schloming.
Repository: qpid-proton-git
Description
-------
If a link is created with the same name as an existing link on the same session, the new link object will never get into the open state, even if the previous link has been closed. This is because in processing the attach, the links on the session are iterated over until one is found with a matching name (and type). However in this case the old link, now in the closed state is returned. While an explicit freeing of the old link before creating the new link avoids this, that is hard to control from e.g. python.
The proposed solution adds an additional check on endpoint state when trying to find the object to whcih an incoming attach refers.
Diffs
-----
proton-c/src/transport/transport.c 62d4742
Diff: https://reviews.apache.org/r/33300/diff/
Testing
-------
All existing tests pass, original reproducers for PRTON-850 and QPID-6320 also pass.
Thanks,
Gordon Sim
Re: Review Request 33300: PROTON-850: allow reusing link names
Posted by Rafael Schloming <rh...@apache.org>.
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/33300/#review80598
-----------------------------------------------------------
Ship it!
I suspect at some point we might need to revisit this area for detach/reattach semantics, but we can cross that bridge when we come to it.
- Rafael Schloming
On April 17, 2015, 11:24 a.m., Gordon Sim wrote:
>
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/33300/
> -----------------------------------------------------------
>
> (Updated April 17, 2015, 11:24 a.m.)
>
>
> Review request for qpid and Rafael Schloming.
>
>
> Repository: qpid-proton-git
>
>
> Description
> -------
>
> If a link is created with the same name as an existing link on the same session, the new link object will never get into the open state, even if the previous link has been closed. This is because in processing the attach, the links on the session are iterated over until one is found with a matching name (and type). However in this case the old link, now in the closed state is returned. While an explicit freeing of the old link before creating the new link avoids this, that is hard to control from e.g. python.
>
> The proposed solution adds an additional check on endpoint state when trying to find the object to whcih an incoming attach refers.
>
>
> Diffs
> -----
>
> proton-c/src/transport/transport.c 62d4742
>
> Diff: https://reviews.apache.org/r/33300/diff/
>
>
> Testing
> -------
>
> All existing tests pass, original reproducers for PRTON-850 and QPID-6320 also pass.
>
>
> Thanks,
>
> Gordon Sim
>
>