Hi Michael and Jielin,
I think this change is low enough risk that we do not require a
specific CTS run. If the updated changes can be provided to me by 3:30
PM EST, I will include them in the CTS run we plan to do tonight.
-Tom
Michael Bouschen wrote:
>Hi Jielin,
>
>a couple of remarks:
>- Please move the test of the left argument type to the block that makes
>sure the left argument is not null. So the test should directly follow
>the line calling left.validateParameter.
>- We need the same test for the right argument (called denominator).
>- We should generalize the error message such that we can reuse it for
>other function nodes. I propose to call the error constant and the
>method invalidFunctionArgument. The message could be:
> Invalid {0} function argument [{1}], expected argument of type [{2}].
>In the case of the MOD function you would call:
> invalidFunctionArgument("MOD", left.getAsString(), "integral type");
>
>Tom,
>do we need a Query CTS run for this change?
>
>Regards Michael
>
>
>
>>Hi, Tom, Michael,
>>
>>Here is the code change for issue 146 - JPQL validation: Parser not
>>catching invalid argument type used with MOD.
>>
>>With more validation in ModNode, for the query
>>
>>select o from Order o WHERE MOD(o.totalPrice, 10) = 5.
>>
>>The EJBQLException will be throwed:
>>
>> [java] Caused by: Exception [TOPLINK-8020] (Oracle TopLink
>>Essentials - 2006.3 (Build 060319)):
>>oracle.toplink.essentials.exceptions.EJBQLException
>> [java] Exception Description: Invalid MOD argument [o.totalPrice],
>>expected an integral data type.
>>
>>Please review the code.
>>
>>Thanks.
>>
>>Jielin
>>
>>
>>
--
Tom Ware
Principal Software Engineer
Oracle Canada Inc.
Direct: (613) 783-4598
Email: tom.ware_at_oracle.com