I noticed update_tests.py 37 has been failing on the Mac buildslave for
some time now. The tests validates that copy-on-update is working for a
file.
The same tests works over ra_svn and ra_local, but fails over ra_neon &
ra_serf on any platform, it's not Mac specific.
So, I looked into the code and found the change in attached patch to fix
the issue.
Current code basically tries to strip of the first slash of the path
found in the repository, but o_path also starts with a slash, so there's
no need to strip of anything.
This looks like a simple patch, but since this code has been on trunk
for a while now (r26471, a month ago) I'm wondering why this issue
hasn't been catched earlier?
Can anyone validate? I've marked the test as XFailing over
ra_[neon/serf] in r26990
thanks,
Lieven
[[[
Fix copy-on-update over ra_[neon/serf]. Don't strip of leading slash when
comparing copy path and target path.
* subversion/libsvn_repos/reporter.c
(add_file_smartly): don't strip of leading slash of closest_copy_path.
]]]
Index: subversion/libsvn_repos/reporter.c
===================================================================
--- subversion/libsvn_repos/reporter.c (revision 26983)
+++ subversion/libsvn_repos/reporter.c (working copy)
@@ -646,7 +646,7 @@
/* If the destination of the copy event is the same path as
o_path, then we've found something interesting that should
have 'copyfrom' history. */
- if (strcmp(closest_copy_path + 1, o_path) == 0)
+ if (strcmp(closest_copy_path, o_path) == 0)
{
SVN_ERR(svn_fs_copied_from(copyfrom_rev, copyfrom_path,
closest_copy_root, closest_copy_path,
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Sat Oct 6 15:46:07 2007