Author: hwright
Date: Fri Nov 20 22:23:07 2009
New Revision: 882738
URL: http://svn.apache.org/viewvc?rev=882738&view=rev
Log:
Recursively lock the working copy to an infinite depth in the event of an
update. (For rationale, see r882730.)
* subversion/libsvn_client/update.c
(svn_client__update_internal): Remove the levels_to_lock calculation, and
just always lock to an infinite depth.
Modified:
subversion/trunk/subversion/libsvn_client/update.c
Modified: subversion/trunk/subversion/libsvn_client/update.c
URL:
http://svn.apache.org/viewvc/subversion/trunk/subversion/libsvn_client/update.c?rev=882738&r1=882737&r2=882738&view=diff
==============================================================================
--- subversion/trunk/subversion/libsvn_client/update.c (original)
+++ subversion/trunk/subversion/libsvn_client/update.c Fri Nov 20 22:23:07 2009
@@ -102,7 +102,6 @@
const char *repos_root;
svn_error_t *err;
svn_revnum_t revnum;
- int levels_to_lock;
svn_wc_traversal_info_t *traversal_info = svn_wc_init_traversal_info(pool);
svn_wc_adm_access_t *adm_access;
svn_boolean_t use_commit_times;
@@ -125,22 +124,6 @@
if (depth == svn_depth_unknown)
depth_is_sticky = FALSE;
- /* ### Ah, the irony. We'd like to base our levels_to_lock on the
- ### depth we're going to use for the update. But that may depend
- ### on the depth in the working copy, which we can't discover
- ### without calling adm_open. We could expend an extra call,
- ### with levels_to_lock=0, to get the real depth (but only if we
- ### need to) and then make the real call... but it's not worth
- ### the complexity right now. If the requested depth tells us to
- ### lock the entire tree when we don't actually need to, that's a
- ### performance hit, but (except for access contention) it is not
- ### a correctness problem. */
-
- /* We may have to crop the subtree if the depth is sticky, so lock the
- entire tree in such a situation*/
- levels_to_lock = depth_is_sticky
- ? -1 : SVN_WC__LEVELS_TO_LOCK_FROM_DEPTH(depth);
-
/* Sanity check. Without this, the update is meaningless. */
SVN_ERR_ASSERT(path);
@@ -157,7 +140,8 @@
*/
SVN_ERR(svn_wc__adm_open_anchor_in_context(&adm_access, &dir_access,
&target, ctx->wc_ctx, path,
- TRUE, levels_to_lock,
+ TRUE,
+ -1, /* recursive lock */
ctx->cancel_func,
ctx->cancel_baton, pool));
}
@@ -166,8 +150,9 @@
/* Assume the exact root is specified (required for externals to work,
as these would otherwise try to open the parent working copy again) */
SVN_ERR(svn_wc__adm_open_in_context(&adm_access, ctx->wc_ctx, path, TRUE,
- levels_to_lock, ctx->cancel_func,
- ctx->cancel_baton, pool));
+ -1, /* recursive lock */
+ ctx->cancel_func, ctx->cancel_baton,
+ pool));
dir_access = adm_access;
target = "";
}
