Philip Martin <philip.martin_at_wandisco.com> writes:
> Another solution would be to have svn_wc__db_read_info return
> status=added for incomplete working nodes so they would be treated as
> added automatically. We would need to have svn_wc__db_scan_addition
> return status=incomplete so that the incomplete status is still
> available. That means that we lose the incomplete information when
> calling read_info, perhaps forcing scan_addition to be used in more
> Index: subversion/libsvn_wc/wc_db.h
> --- subversion/libsvn_wc/wc_db.h (revision 1177632)
> +++ subversion/libsvn_wc/wc_db.h (working copy)
> @@ -1710,8 +1710,8 @@
> node's revision.
> - The BASE or WORKING node is incomplete due to an interrupted
> - operation.
> + The BASE is incomplete due to an interrupted operation. An
> + incomplete WORKING node will be svn_wc__db_status_added.
> If REVISION is requested, it will be set to the revision of the
> unmodified (BASE) node, or to SVN_INVALID_REVNUM if any structural
> @@ -2441,6 +2441,8 @@
> ancestor unshadowed BASE node. ORIGINAL_* will indicate the source
> of the copy.
> + svn_wc__db_status_incomplete -- this NODE is copied but incomplete.
> svn_wc__db_status_moved_here -- this NODE arrived as a result of a move.
> The root of the moved nodes will be stored in OP_ROOT_ABSPATH.
> Similar to the copied state, its parent may be a WORKING node or a
This is looking like the better solution. I think it is the better fit
for the read_info/scan_addition API, and as evidence the patch I
developed for issue 4025 also fixes issue 4026.
uberSVN: Apache Subversion Made Easy
Received on 2011-09-30 17:02:27 CEST