Re: Problem with svn_wc_process_committed

Justin Erenkrantz <jerenkrantz@apache.org> writes:

> Prior to r2470, we used to unlock the adm_access 'baton' before
> calling svn_wc_process_committed in svn_client_commit(). In r2470,
> we now delay the unlock until after svn_wc_process_committed is
> called. That doesn't work because svn_wc_process_committed doesn't
> take the hash table of locked directories. Therefore, it tries to
> open a lock on a directory that was already held in locked_dirs.
> And, we now fail commit test #8 with my caching patch because we
> already have a lock on the subdirs.

Does your cacheing patch change the locking in some way? I don't see
a test failure with unmodified r2478.

svn_wc_process_committed only locks children so the failure you are
seeing appears to imply that commit is trying to commit items that are
in an ancestor-child relationship. I though the commit processing
stripped out the children in this case, prior to taking out locks.

-- 
Philip Martin
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org