On Sat, Jun 19, 2010 at 01:47, Senthil Kumaran S <sent...@collab.net> wrote: > Greg Stein wrote: >> On Fri, Jun 18, 2010 at 02:39, <style...@apache.org> wrote: >>> ... >>> +++ subversion/trunk/subversion/libsvn_wc/update_editor.c Fri Jun 18 >>> 06:39:31 2010 >>> @@ -5208,10 +5208,12 @@ make_editor(svn_revnum_t *target_revisio >>> if (depth == svn_depth_unknown) >>> depth_is_sticky = FALSE; >>> >>> - /* Get the anchor entry, so we can fetch the repository root. */ >>> - SVN_ERR(svn_wc__node_get_repos_info(&repos_root, &repos_uuid, wc_ctx, >>> - anchor_abspath, TRUE, FALSE, >>> - result_pool, scratch_pool)); >>> + /* Get the anchor's repository root and uuid. */ >>> + SVN_ERR(svn_wc__db_read_info(NULL,NULL, NULL, NULL, &repos_root, >>> &repos_uuid, >>> + NULL, NULL, NULL, NULL, NULL, NULL, NULL, >>> NULL, >>> + NULL, NULL, NULL, NULL, NULL, NULL, NULL, >>> NULL, >>> + NULL, NULL, wc_ctx->db, anchor_abspath, >>> + result_pool, scratch_pool)); >> >> These are not equivalent. The read_info() might return NULL values for >> repos_*, indicating they are inherited from the parent. The node >> function will grab the data from the parent, if/when necessary. > > What will be the alternative function to use here?
Stick to the node function, or go look at its contents. The point is: if NULL is returned, then you need to scan upwards for the repository information. The other thing to note here: I believe that the BASE node *must* exist if you're going to run an update on it. Thus, read_info() is a bit of overkill, and you can simply use base_get_info() and (if needed) scan_base_repos(). I'd suggest this latter approach, if possible. Cheers, -g
