Re: svn commit: rev 3709 - in trunk/subversion: libsvn_wc libsvn_client

Greg Hudson <ghudson@MIT.EDU> writes:

> On Sat, 2002-11-09 at 08:47, philip@tigris.org wrote:
> > (temp_file_plain_cleanup_handler): Use parent pool.
>
> Is that okay, if in actuality it is the parent pool (or its parent,
> etc.) being cleaned up?

Good point :-( It will happen earlier in the process, because pool
cleanup cleans subpools before running cleanup handlers. So the
parent pool will be cleaning subpools, while the subpool is running
cleanup handlers. In practice it is enough to fix the bug we were
seeing.

The thing that worries me, something I have just recently thought of,
is that the parent pool could be the global pool. If that were the
case all the thread-safety problems would reappear.

We could attempt extend our existing svn_pool_create/clear/destroy
with an svn_pool_register_cleanup and maintain an application level
set of cleanup handlers which run before APR starts to cleanup. This
way we get full use of the pool in the cleanup handler. We would also
need to maintain lists of subpools for each pool, so that we could run
cleanup on the subpools. Duplicating APR in this way doesn't sound
like a good idea.

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