Re: svn commit: r22765 - trunk/subversion/libsvn_client

On Wed, 20 Dec 2006, Kamesh Jayachandran wrote:

> dlr@tigris.org wrote:
> >Modified:
> > trunk/subversion/libsvn_client/copy.c
> >
> >Modified: trunk/subversion/libsvn_client/copy.c
> >URL:
> >http://svn.collab.net/viewvc/svn/trunk/subversion/libsvn_client/copy.c?pathrev=22765&r1=22764&r2=22765
> >==============================================================================
> >--- trunk/subversion/libsvn_client/copy.c (original)
> >+++ trunk/subversion/libsvn_client/copy.c Wed Dec 20 06:06:12 2006
> >
> > /* Check that all of our SRCs exist, and all the DSTs don't. */
> > for (i = 0; i < copy_pairs->nelts; i++)
> >@@ -221,7 +220,7 @@
> > ((svn_client__copy_pair_t **) (copy_pairs->elts))[i];
> > svn_node_kind_t dst_kind, dst_parent_kind;
> >
> >- svn_pool_clear(subpool);
> >+ svn_pool_clear(iterpool);
> >
> > /* Verify that SRC_PATH exists. */
> > SVN_ERR(svn_io_check_path(pair->src, &pair->src_kind, pool));
>
> Why not this svn_io_check_path from iterpool?

As I mentioned in a separate email, this operation allocates memory
for our input parameters. I want to avoid altering their values, then
destroying the memory holding their new values. While I don't believe
that those values are used now, this is a safe practice which leaves
the code without booby-traps (and thus easier to modify).

 - Dan