Re: inconsistent rados semantics between automake/cmake builds

[Casey Bodley <cbodley@redhat.com>]

Hi Noah,

We were tracking this bug in rgw for a while, and couldn't find a
simple reproducer. Your test case made it so much easier to fix,
thank you!

A pull request with the fix is pending at:
https://github.com/ceph/ceph/pull/7328

Casey

----- Original Message -----
> On 21/01/2016, Noah Watkins wrote:
> > I'm seeing an anomaly of semantics between Ceph built with Automake
> > and with CMake. I've not been lucky so far tracking this down so I
> > thought I'd toss it out here.
> 
> All right.  Commit 9717714cf62eeb7db979b0e6b509ef305cb757eb in
> github.com/cohortfsllc/cohort-ceph should have this. It's the top entry on
> the
> mtime branch in that repository. If you like things web-style,
> https://github.com/cohortfsllc/cohortfs-ceph/commit/9717714cf62eeb7db979b0e6b509ef305cb757eb
> 
> Casey chased this down to the encode/decode functions and once I took a look
> at
> them I realized the cast was screwy. It now passes your unit test reliably
> (at
> least it worked a few times) under cmake.
> 
> --
> Senior Software Engineer           Red Hat Storage, Ann Arbor, MI, US
> IRC: Aemerson@{RedHat, OFTC, Freenode}
> 0x80F7544B90EDBFB9 E707 86BA 0C1B 62CC 152C  7C12 80F7 544B 90ED BFB9
>