At Tue, 14 May 2013 15:28:16 +0800, Liu Yuan wrote: > > On 05/14/2013 03:22 PM, MORITA Kazutaka wrote: > > We > > cannot cache sha1 digests of stale objects blindly because they may be > > hard-linked to the writable objects in the working directory. > > Why? We use reanme(2) to stale objects, isn't it atomic operation?
We use link(2) to recover objects from local. Then the stale object can be writable again. Anyway, I wonder if it is important to take the stale objects into account. They will be removed just after sheep finishes object recovery, so sheep doesn't have so many chances to use the cached sha1 digests. Thanks, Kazutaka -- sheepdog mailing list [email protected] http://lists.wpkg.org/mailman/listinfo/sheepdog
