Re: svn commit: r22610 - in trunk/subversion: include libsvn_fs libsvn_fs_base libsvn_fs_fs libsvn_repos mod_dav_svn tests/libsvn_fs

Hi Mike,

On Fri, Dec 08, 2006 at 08:16:31AM -0800, cmpilato@tigris.org wrote:
> * subversion/libsvn_repos/replay.c
> (svn_repos_replay2): use new svn_fs_txn_root_base_revision() instead
> of the roundabout way.
>

> Modified: trunk/subversion/libsvn_repos/replay.c
> URL: http://svn.collab.net/viewvc/svn/trunk/subversion/libsvn_repos/replay.c?pathrev=22610&r1=22609&r2=22610
> ==============================================================================
> --- trunk/subversion/libsvn_repos/replay.c (original)
> +++ trunk/subversion/libsvn_repos/replay.c Fri Dec 8 08:16:30 2006
> @@ -711,26 +711,20 @@
>
> if (send_deltas)
> {
> + svn_revnum_t compare_root_revision = SVN_INVALID_REVNUM;
> + if (svn_fs_is_revision_root(root))
> + compare_root_revision = svn_fs_revision_root_revision(root) - 1;
> + else if (svn_fs_is_txn_root(root))
> + compare_root_revision = svn_fs_txn_root_base_revision(root);

Isn't it either one or the other? The old code didn't do the second
check -- do you think we shouldn't be assuming that there are only two
types of root?

Regards,
Malcolm