On Monday 22 October 2012 06:53 PM, Hyrum K Wright wrote:
On Mon, Oct 22, 2012 at 7:55 AM, Philip Martin
<philip.mar...@wandisco.com> wrote:
hwri...@apache.org writes:
Author: hwright
Date: Sun Oct 21 01:18:26 2012
New Revision: 1400545
URL: http://svn.apache.org/viewvc?rev=1400545&view=rev
Log:
Refactor the pre-1.5 fallback code for replay_ranges into a separate function.
* subversion/libsvn_ra/ra_loader.c
(replay_range_from_replays): New.
(svn_ra_replay_range): Call the factored out code.
This causes svnrdump_tests.py 43 and lots of svnsync_tests.py to FAIL
over ra_svn. It looks like an innocuous change, I can't see the error
in the code.
Thanks for pointing out the failures. I won't be able to get around
to fixing it for (at least) several hours, so feel free to revert in
the meantime if it's a large issue.
If it is a serious issue, the attached patch will fix it. But I am
trying to understand what is wrong with the previous commit.
Thanks & Regards,
Vijayaguru
-Hyrum
Index: subversion/libsvn_ra/ra_loader.c
===================================================================
--- subversion/libsvn_ra/ra_loader.c (revision 1400760)
+++ subversion/libsvn_ra/ra_loader.c (working copy)
@@ -1189,17 +1189,19 @@
revstart_func, revfinish_func,
replay_baton, pool);
- if (err && (err->apr_err != SVN_ERR_RA_NOT_IMPLEMENTED))
- return svn_error_trace(err);
+ if (err && (err->apr_err == SVN_ERR_RA_NOT_IMPLEMENTED))
+ {
+ svn_error_clear(err);
+ return svn_error_trace(replay_range_from_replays(session, start_revision,
+ end_revision,
+ low_water_mark,
+ text_deltas,
+ revstart_func,
+ revfinish_func,
+ replay_baton, pool));
- svn_error_clear(err);
- return svn_error_trace(replay_range_from_replays(session, start_revision,
- end_revision,
- low_water_mark,
- text_deltas,
- revstart_func,
- revfinish_func,
- replay_baton, pool));
+ }
+ return svn_error_trace(err);
}
svn_error_t *