On Sat, Oct 31, 2009 at 07:04:57AM -0400, Greg Stein wrote:
> On Thu, Oct 29, 2009 at 19:59, Paul T. Burba <pburba_at_collab.net> wrote:
> >...
> > +++ trunk/subversion/libsvn_wc/copy.c Thu Oct 29 16:59:24 2009 (r40306)
> >...
> > @@ -355,13 +294,36 @@ determine_copyfrom_info(const char **cop
> > src_copyfrom_relpath, result_pool);
> > rev = src_copyfrom_rev;
> > }
> > - else
> > + else if (status == svn_wc__db_status_added
> > + || status == svn_wc__db_status_obstructed_add)
> > {
> > /* ...But if this file is merely the descendant of an explicitly
> > copied/moved directory, we need to do a bit more work to
> > determine copyfrom_url and copyfrom_rev. */
> > - SVN_ERR(get_copyfrom_url_rev_via_parent(&url, &rev, db, src_abspath,
> > - scratch_pool, scratch_pool));
> > + const char *op_root_abspath;
> > + const char *original_repos_relpath;
> > + const char *original_root_url;
> > + svn_revnum_t original_revision;
> > + const char *op_root_rel_path;
> > +
> > + SVN_ERR(svn_wc__db_scan_addition(&status, &op_root_abspath,
> > + NULL, NULL, NULL,
> > + &original_repos_relpath,
> > + &original_root_url, NULL,
> > + &original_revision,
>
> Why not directly query into &rev ?
>
> > + db, src_abspath,
> > + scratch_pool, scratch_pool));
> > + url = svn_uri_join(original_root_url, original_repos_relpath,
> > + result_pool);
> > + op_root_rel_path = svn_dirent_is_child(op_root_abspath, src_abspath,
> > + scratch_pool);
> > + url = svn_uri_join(url, op_root_rel_path, scratch_pool);
> > + rev = original_revision;
> > + }
>
> url is probably going into the wrong pool here. The first join should
> definitely use scratch_pool, and I'm guessing the second was probably
> intended for result_pool.
No, because the joined result is later dupped into the result pool.
Your concerns should be addressed by r4032{1,2,3}.
Stefan
------------------------------------------------------
http://subversion.tigris.org/ds/viewMessage.do?dsForumId=462&dsMessageId=2413225
Received on 2009-10-31 13:33:33 CET