[email protected] wrote:
Author: hwright
Date: Tue Nov 24 16:25:16 2009
New Revision: 883765
URL: http://svn.apache.org/viewvc?rev=883765&view=rev
Log:
Make fs-pack-test 5 pass by attempting to refresh the min-unpacked-revprop
value in the event of an initial failure.
* subversion/tests/libsvn_fs_fs/fs-pack-test.c
(test_funcs): Mark get_set_revprop_packed_fs() as PASS.
* subversion/libsvn_fs_fs/fs_fs.c
(revision_proplist): New, copied from svn_fs_fs__revision_proplist().
(svn_fs_fs__revision_proplist): Implement as a wrapper around the above,
potentially reloading the min-unpacked-revprop value as needed.
+svn_error_t *
+svn_fs_fs__revision_proplist(apr_hash_t **proplist_p,
+ svn_fs_t *fs,
+ svn_revnum_t rev,
+ apr_pool_t *pool)
+{
+ svn_error_t *err;
+
+ err = revision_proplist(proplist_p, fs, rev, pool);
+ if (err && err->apr_err == SVN_ERR_FS_NO_SUCH_REVISION)
I haven't thought too closely about this race, but could this happen more than
once? Should there be a while loop around revision_proplist() instead checking
for SVN_ERR_FS_NO_SUCH_REVISION?
Blair
