Question about lang/wisconsin.sql test

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

lang/wisconsin.sql test currenly does not work in J2ME environment
because the test setup uses stored procedures with nested connection.
To make this test run in this environment, I am planning to move the
setup (sql for table creation and inserts) to a Java class
(Wisconsin).
>From the Java class, run the remaining sql using ij and redirect the
output which can be used to compare with the master file.

Since this is a benchmark test, I would like to know if this approach
is okay and does not compromise any testing. Please let me know if
anyone thinks this is not okay.

Thanks,
Deepa

Re: Question about lang/wisconsin.sql test

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

Thanks Myrna. I have attached a patch for this test to DERBY-398.

Deepa

Re: Question about lang/wisconsin.sql test

[Myrna van Lunteren <m....@gmail.com>]

Hi Deepa,

This test, as it runs in the function test harness, is not actually running
as a performance benchmark. If I remember correctly, it was thought to add
value in the functional suite because there were interesting aspects of
creating tables and executing queries of this type and gathering the
statistics. I probably used the easiest way (least amount of work for me) I
could think of converting it from the actual performance benchmark we had
running at the time as a performance test. Changing the setup mechanism will
be ok as long as it doesn't impact the functional / optimizer behavior.
If I'm off on this someone please correct me...

Myrna

On 12/28/05, Deepa Remesh <dr...@gmail.com> wrote:
>
> lang/wisconsin.sql test currenly does not work in J2ME environment
> because the test setup uses stored procedures with nested connection.
> To make this test run in this environment, I am planning to move the
> setup (sql for table creation and inserts) to a Java class
> (Wisconsin).
> From the Java class, run the remaining sql using ij and redirect the
> output which can be used to compare with the master file.
>
> Since this is a benchmark test, I would like to know if this approach
> is okay and does not compromise any testing. Please let me know if
> anyone thinks this is not okay.
>
> Thanks,
> Deepa
>