Some more (expected?) failures in tests with 10.1 client/10.2 server

[Deepa Remesh <dr...@gmail.com>]

Seeing these additional failures in the client/server compat tests.
These seem to be expected but just wanted to get feedback from others
in case we need to do something extra:

* lang/forupdate.sql (after DERBY-1361)
* lang/holdCursorIJ.sql (after DERBY-1361)
* jdbcapi/bestrowidentifier.sql (after DERBY-137)
* jdbcapi/metadataJdbc20.java (after DERBY-137)
* derbynet/dblook_test_net.java (after DERBY-1087)
* derbynet/testProtocol.java (after DERBY-959)

Sample diffs are:

*** Start: forupdate jdk1.4.2 DerbyNetClient derbynetmats:derbynetmats
2006-06-29 12:41:52 ***
170 del
< ERROR XCL08: Cursor 'SQL_CURLH000C1' is not on a row.
170a170
> ERROR 24000: Invalid cursor state - no current row.
Test Failed.
*** End:   forupdate jdk1.4.2 DerbyNetClient derbynetmats:derbynetmats
2006-06-29 12:42:07 ***

*** Start: bestrowidentifier jdk1.4.2 DerbyNetClient
derbynetmats:jdbcapi 2006-06-29 12:52:33 ***
25 del
< SCOPE |COLUMN_NAME |DATA_& |TYPE_NAME |COLUMN_SIZE |BUFFER_LEN&
|DECIM& |PSEUD&
25a25
> SCOPE |COLUMN_NAME |DATA_TYPE |TYPE_NAME |COLUMN_SIZE |BUFFER_LEN& |DECIM& |PSEUD&

*** Start: metadataJdbc20 jdk1.4.2 DerbyNetClient derbynetmats:jdbc20
2006-06-29 13:03:28 ***
5 del
< TYPE_CAT,TYPE_SCHEM,TYPE_NAME,CLASS_NAME,DATA_TYPE,REMARKS
5a5
> TYPE_CAT,TYPE_SCHEM,TYPE_NAME,CLASS_NAME,DATA_TYPE,REMARKS,BASE_TYPE

*** Start: dblook_test_net jdk1.4.2 DerbyNetClient
derbynetmats:derbynetmats 2006-06-29 11:33:28 ***
1335,1340d1334
< <StoredPreparedStatement>APP.<sysname> -> <Conglomerate>BAR.<sysname>
< StoredPreparedStatement
< <systemid>
< Conglomerate
< -----
< <systemid>
1483a1478,1489
> <systemid>
> <StoredPreparedStatement>APP.<sysname> -> <Conglomerate>BAR.<sysname>
> StoredPreparedStatement
> <systemid>
> Conglomerate
> -----
> <systemid>
> <StoredPreparedStatement>APP.<sysname> -> <Conglomerate>BAR.<sysname>
> StoredPreparedStatement
> <systemid>
> Conglomerate
> -----
Test Failed.
*** End:   dblook_test_net jdk1.4.2 DerbyNetClient
derbynetmats:derbynetmats 2006-06-29 11:34:07 ***

Thanks,
Deepa

Re: Some more (expected?) failures in tests with 10.1 client/10.2 server

[Andreas Korneliussen <An...@Sun.COM>]

Deepa Remesh wrote:
> Seeing these additional failures in the client/server compat tests.
> These seem to be expected but just wanted to get feedback from others
> in case we need to do something extra:
> 
> * lang/forupdate.sql (after DERBY-1361)
> * lang/holdCursorIJ.sql (after DERBY-1361)

The change in DERBY-1361 is known to give another error message/SQLState 
  when trying to do an update after a commit, so the failures are expected.

< ERROR XCL08: Cursor 'SQL_CURLH000C1' is not on a row.
> ERROR 24000: Invalid cursor state - no current row.

Andreas

Re: Some more (expected?) failures in tests with 10.1 client/10.2 server

[Sunitha Kambhampati <ks...@gmail.com>]

Deepa Remesh wrote:

> Seeing these additional failures in the client/server compat tests.
> These seem to be expected but just wanted to get feedback from others
> in case we need to do something extra:
>
> * derbynet/testProtocol.java (after DERBY-959)

The max query blocksize that is allowed in the server has changed with 
derby-959 from 32k to 10M.  The 10.1 protocol.tests file has tests for 
invalid query block size and sends a value of 40000 and checks for the 
syntax error.  But after 959 changes, the server will not throw an 
exception for 40000 value and thus the test fails.  The diffs related to 
invalid query block size and too large value for qryblksz are expected.

There is no compatibility issue with this particular change.
---------------------

Even without the 959 changes, I patched the 10.2 server with Knut's 
patch for DERBY-1425 and there are diffs with the testProtocol.java test.
I found Knut's comments on DERBY-1425 interesting with respect to  
testProtocol.java & compatibility testing
http://issues.apache.org/jira/browse/DERBY-1425#action_12417731

Thanks,
Sunitha.

Re: Some more (expected?) failures in tests with 10.1 client/10.2 server

[Knut Anders Hatlen <Kn...@Sun.COM>]

Deepa Remesh <dr...@gmail.com> writes:

> * jdbcapi/bestrowidentifier.sql (after DERBY-137)
> * jdbcapi/metadataJdbc20.java (after DERBY-137)

These two are expected and believed to be compatible (new columns at
the end of metadata result sets, and one column changed type from
short to int).

-- 
Knut Anders