You are viewing a plain text version of this content. The canonical link for it is here.
Posted to derby-dev@db.apache.org by "Mike Matrigali (JIRA)" <de...@db.apache.org> on 2006/10/09 17:44:20 UTC
[jira] Resolved: (DERBY-1939) ERROR XSDA7 while executing query
that spills to BackingStoreHashTable
[ http://issues.apache.org/jira/browse/DERBY-1939?page=all ]
Mike Matrigali resolved DERBY-1939.
-----------------------------------
Fix Version/s: 10.1.3.2
Resolution: Fixed
committed fix to 10.1, also checked in new test for this problem. The problem is already fixed in 10.2.
Fixes HeapRowLocation.getNewNull() to return actual null object, this
functionality is used by the hash overflow to disk code. Without this
change the HeapRowLocation is correctly written out to the StoredPage,
but on read the overflow code passes a null template which causes the
wrong code to read the data back into the object.
> ERROR XSDA7 while executing query that spills to BackingStoreHashTable
> ----------------------------------------------------------------------
>
> Key: DERBY-1939
> URL: http://issues.apache.org/jira/browse/DERBY-1939
> Project: Derby
> Issue Type: Bug
> Components: Store
> Affects Versions: 10.1.3.1, 10.1.3.2
> Reporter: A B
> Assigned To: Mike Matrigali
> Fix For: 10.1.3.2
>
> Attachments: d1939.java, d1939_log.txt
>
>
> I have a database with a large query that's failing in Derby 10.1 with:
> ERROR XSDA7: Restore of a serializable or SQLData object of class , attempted to read more data than was
> originally stored
> The error only occurs if I use a PreparedStatement for the query and if I specify a parameter marker for one particular equality operand. If I replace the parameter with an actual value, the query runs without error.
> I was eventually able to come up with a standalone repro that uses nonsense tables and a nonsense query but still demonstrates the problem. That said, the error does *not* occur in Derby 10.2 or main, so it has apparently been fixed in those codelines.
> I did some investigating and was eventually able to track the fix down to svn commit # 389202 (DERBY-690). Before that checkin the query fails against 10.2, but after that it runs without error. It turns out that there's one uncommented change in svn 389202 that seems to have fixed the problem, namely:
> Index: java/engine/org/apache/derby/impl/store/access/heap/HeapRowLocation.java
> ===================================================================
> --- java/engine/org/apache/derby/impl/store/access/heap/HeapRowLocation.java (revision 389043)
> +++ java/engine/org/apache/derby/impl/store/access/heap/HeapRowLocation.java (working copy)
> @@ -104,7 +104,7 @@
> }
>
> public DataValueDescriptor getNewNull() {
> - return null;
> + return new HeapRowLocation();
> }
>
> To confirm I undid this one line change in the 10.3 codeline and then the query started failing there. Similarly, after I made this one-line change in the latest 10.1 codeline, the query started running successfully. So apparently this is the fix to the problem...though I don't really know why...?
--
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators: http://issues.apache.org/jira/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira