On Sep 12, 2009, at 4:11 AM, Greg Stein wrote:
...
>> ...
>> @@ -1471,27 +1478,27 @@ modcheck_found_entry(const char *path,
>> static svn_error_t *
>> tree_has_local_mods(svn_boolean_t *modified,
>> svn_boolean_t *all_edits_are_deletes,
>> - const char *path,
>> - svn_wc_adm_access_t *adm_access,
>> + svn_wc__db_t *db,
>> + const char *local_abspath,
>> svn_cancel_func_t cancel_func,
>> void *cancel_baton,
>> apr_pool_t *pool)
>> {
>> - static const svn_wc_entry_callbacks2_t modcheck_callbacks =
>> - { modcheck_found_entry, svn_wc__walker_default_error_handler };
>> + static const svn_wc__node_walk_callbacks_t modcheck_callbacks =
>> + { modcheck_found_node, svn_wc__walker_default_error_handler };
>
> I didn't review (yet) the new node walker, but an error callback is
> bogus in the extreme. Thus, a struct of callbacks is wrong. Just pass
> a simple function. *IF* a caller wants special error handling, then
> they can do it *before* returning an error to the walker. There is NO
> reason for the walker to be involved. I looked at this once, and
> (iirc) the usage is a single occurrence. Totally bogus,
> over-engineered walker blather. Please don't repeat it with the node
> walker.
Removed the "totally bogus, over-engineered walker blather" in r39259.
-Hyrum
------------------------------------------------------
http://subversion.tigris.org/ds/viewMessage.do?dsForumId=462&dsMessageId=2394155
Received on 2009-09-13 04:37:00 CEST