You don't have commiter privileges?!?!
> -----Original Message-----
> From: Gianny Damour [mailto:[EMAIL PROTECTED]
> Sent: Sunday, October 03, 2004 8:00 AM
> To: [EMAIL PROTECTED]
> Cc: Jeremy Boynes
> Subject: [openejb-dev] CMR support + compound PK - patch
>
> Hi,
>
>
> I am profiting from this slow-down of the updates to the deployment
code
> to refresh these two patches:
> * http://jira.codehaus.org/browse/TQL-3; and
> * http://jira.codehaus.org/browse/OPENEJB-13.
>
> They enable the CMR capabilities of TranQL and complete the support of
> compound PK.
>
> During the implementation of these two patches, I have found two
issues,
> that I have deliberately left aside for this second pass (the first
was
> CMR only without compound PK):
> - ejbFind<METHOD> do not add to the InTxCache CacheRow corresponding
to
> the instances, which have been found. This means that the following
code
> do not work:
>
> <start the transaction>
> CompoundPK pkA = new CompoundPK(new Integer(2), "value2");
> ALocal a = ahome.create(pkA);
> BLocal b = bhome.findByPrimaryKey(new Integer(11));
> a.setB(b);
> <commit transaction>
>
> They do not work as no CacheRow corresponding to b is in the
InTxCache.
> I have avoided this issue by performing a dummy load operation (e.g.
> b.getField1()) just after the findByPrimaryKey call. So, if you check
> the various TestCase, you will see some of these weird lines.
>
> I think that ejbFind<Method> should add to the InTxCache the relevant
> CacheRow and also inject the identity into this CacheRow.
>
> - an Axion bug causes various tests to fail. Basically, the following
> update does not work as expected:
>
> // The following PrepareStatement does not set to null fka
> // PreparedStatement ps = null;
> // ps = c.prepareStatement("UPDATE B SET value = CASE WHEN ? THEN
?
> ELSE value END, fka = CASE WHEN ? THEN ? ELSE fka END WHERE b1 = ?");
> // ps.setBoolean(1, false);
> // ps.setString(2, "");
> // ps.setBoolean(3, true);
> // ps.setNull(4);
> // ps.setInt(5, 1);
> // ps.execute();
>
> Due to this reason, some tests have been disabled. Though, they have
> been successfully executed with the DataSource of MySQL.
>
>
> At this stage, CMP EntityBean with unknown PK are not supported. I
think
> that this is the "last" limitation, which stops us to deploy PetStore
> 1.3. I will start working on that on Tuesday. Hopefully, it should be
> completed by the end of next week.
>
> Thanks,
> Gianny
