Re: svn commit: r38848 - trunk/subversion/libsvn_client

Stsp,

so you're not able to reproduce this error anymore, apparently. Can this
comment be removed again? (see below)

~Neels

Stefan Sperling wrote:
> Author: stsp
> Date: Wed Aug 19 07:43:12 2009
> New Revision: 38848
>
> Log:
> * subversion/libsvn_client/repos_diff.c
> (add_directory): Add a temporary comment regarding a failing merge test.
>
> Modified:
> trunk/subversion/libsvn_client/repos_diff.c
>
> Modified: trunk/subversion/libsvn_client/repos_diff.c
> URL: http://svn.collab.net/viewvc/svn/trunk/subversion/libsvn_client/repos_diff.c?pathrev=38848&r1=38847&r2=38848
> ==============================================================================
> --- trunk/subversion/libsvn_client/repos_diff.c Wed Aug 19 06:43:21 2009 (r38847)
> +++ trunk/subversion/libsvn_client/repos_diff.c Wed Aug 19 07:43:12 2009 (r38848)
> @@ -936,6 +936,13 @@ add_directory(const char *path,
> this is a tree-conflict case. The path is already marked tree-
> conflicted (either by some previous run altogether, or by this
> replace's delete operation). No need to notify again here. */
> + /* ### Failing test merge_test 132:
> + * ### There is nothing that guarantees that the delete will come
> + * ### before the add. If you trace the merge done in merge_test 132
> + * ### and break at subversion/libsvn_client/merge.c:merge_file_added
> + * ### and subversion/libsvn_client/merge.c:merge_file_deleted, you
> + * ### can see that the add gets called first, then the delete (for
> + * ### the file "mu"). */
> svn_wc_notify_t *notify;
> svn_boolean_t is_replace = FALSE;
> deleted_path_notify_t *dpn = apr_hash_get(eb->deleted_paths, b->wcpath,
>
> ------------------------------------------------------
> http://subversion.tigris.org/ds/viewMessage.do?dsForumId=495&dsMessageId=2385252

------------------------------------------------------
http://subversion.tigris.org/ds/viewMessage.do?dsForumId=462&dsMessageId=2386178