Re: svn commit: r979891 - in /apr/apr/trunk: build/ dbd/ dbm/ file_io/netware/ file_io/os2/ file_io/unix/ file_io/win32/ memory/unix/ threadproc/beos/ threadproc/unix/

[Stefan Fritsch]

On Wed, 28 Jul 2010, Ruediger Pluem wrote:
On 07/28/2010 12:09 AM, s...@apache.org wrote:
Author: sf
Date: Tue Jul 27 22:09:45 2010
New Revision: 979891

Modified: apr/apr/trunk/dbm/apr_dbm_sdbm.c
URL: 
http://svn.apache.org/viewvc/apr/apr/trunk/dbm/apr_dbm_sdbm.c?rev=979891&r1=979890&r2=979891&view=diff
==============================================================================
--- apr/apr/trunk/dbm/apr_dbm_sdbm.c (original)
+++ apr/apr/trunk/dbm/apr_dbm_sdbm.c Tue Jul 27 22:09:45 2010
@@ -193,7 +193,7 @@ static apr_status_t vt_sdbm_nextkey(apr_
     pkey->dsize = rd.dsize;

     /* store any error info into DBM, and return a status code. */
-    return set_error(dbm, APR_SUCCESS);
+    return set_error(dbm, rv);

This breaks the testdbm test of apr.

Sorry, reverted in r980197 for now.

This discards any error from getnext() in sdbm.c, including those from the 
chkpage consistency check. But ATM, I am not sure how to fix this 
correctly.