Re: Problem with svn_wc_process_committed

Justin Erenkrantz <jerenkrantz@apache.org> writes:

> On Fri, Jul 12, 2002 at 07:16:34PM +0100, Philip Martin wrote:
> > But svn_path_condense_targets is called before harvest_committables
> > and that is supposed to remove the ancestor/descendant relationship.
>
> No, but harvest_committables *needs* to walk down the tree and read
> every entries file for its descendants. So, it'd have to lock the
> administrative directories as it determines what to commit.
>
> > > Yes, that is what commit test #8 does. I don't believe the commit
> > > processing strips out children. -- justin
> >
> > I'm still confused, as it doesn't fail when I run the test.
>
> Because harvest_committables isn't locking anything now. I said
> before that you need to apply my caching patch. I wanted to check
> in the code that would pass the baton around and I can't do that
> because svn_wc_process_committed is broken. -- justin

Ah, I see, I didn't twig that you were putting more locks into the
code, sorry.

Broken is a bit strong, I said I would introduce the new locks in
stages. I'll just claim that your code violates the current
pre-conditions for calling svn_wc_process_committed :-)

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