Stuff like this, and mismatched symbols between the name/len are why I
switched most of the apr_hash_[gs]et() to use APR_HASH_KEY_STRING. If we
keep that pattern, then we could reduce "thinko" / typo bugs.
Considering the quantity of string keys that we have, I even considered
adding apr_hash_gets() and apr_hash_sets() calls. (or maybe getstr/setstr)
But those are a bit of a misnomer since we aren't getting/setting a
string... we're just using a string as a key.
Cheers,
-g
On Mon, Oct 16, 2000 at 10:22:34PM -0000, sussman@tigris.org wrote:
> User: sussman
> Date: 00/10/16 15:22:34
>
> Modified: subversion/libsvn_wc adm_crawler.c
> Log:
> (process_subdirectory): fix silly recursion bug.
>
> Revision Changes Path
> 1.30 +2 -3 subversion/subversion/libsvn_wc/adm_crawler.c
>
> Index: adm_crawler.c
> ===================================================================
> RCS file: /cvs/subversion/subversion/libsvn_wc/adm_crawler.c,v
> retrieving revision 1.29
> retrieving revision 1.30
> diff -u -r1.29 -r1.30
> --- adm_crawler.c 2000/10/16 21:57:04 1.29
> +++ adm_crawler.c 2000/10/16 22:22:34 1.30
> @@ -402,8 +402,7 @@
>
> /* Grab the entry representing "." */
> this_dir = (svn_wc__entry_t *)
> - apr_hash_get (entries, SVN_WC__ENTRIES_THIS_DIR,
> - sizeof(SVN_WC__ENTRIES_THIS_DIR));
> + apr_hash_get (entries, SVN_WC__ENTRIES_THIS_DIR, APR_HASH_KEY_STRING);
> if (! this_dir)
> return
> svn_error_createf (SVN_ERR_WC_ENTRY_NOT_FOUND, 0, NULL, subpool,
> @@ -656,7 +655,7 @@
> /* Recurse, using a NULL dir_baton. Why NULL? Because that
> will later force a call to do_dir_replaces() and get the
> _correct_ dir baton for the child directory. */
> - err = process_subdirectory (path, NULL,
> + err = process_subdirectory (full_path_to_entry, NULL,
> editor, edit_baton,
> stack, filehash, subpool);
> if (err) return err;
>
>
>
--
Greg Stein, http://www.lyra.org/
Received on Sat Oct 21 14:36:11 2006