On Thu, 16 Sep 2004 14:47:34 -0500, Jason Hildebrand wrote:
> On Sat, 2004-09-04 at 13:15, Winston WOLFF wrote:
>> I have some patches for the PostgreSQL support in MiddleKit. I
>> know what the testing procedure is around here, so I am hesitant
>> to check them in actuallly.
> Hi Winston,
> Thanks for the patches! Generally it's best to post the patches on
> the SourceForge project page, to make sure they don't get lost.
> Also, please use the unified diff ("diff -u" or "cvs diff -u") when
> generating patches; they're easier to read and apply more easily.
> I've applied these patches and committed them to CVS.
>> Also for the last patch in SQLGenerator.py
>> will break support for MySQL and MSSQL, but I don't know enough
>> about adding in logic to differentiate it.
> I split out the "Connect to database" command into a new method
> called writeConnectToDatabase in SQLGenerator.py, so that it can be
> overridden in PostgreSQLSQLGenerator.py. This solves the problem
> for Postgres without changing behaviour for the other databases.
> It takes a bit to understand the use of mixins in MiddleKit so that
> you know where and how to override things -- it's unfortunately not
> immediately obvious how to get stuff done.
> I also made a few other PostgreSQL fixes to make the tests pass for
> MKString and MKDateTime, and now the test suite produces:
> succeeded MKBasic.mkmodel
> succeeded MKNone.mkmodel
> succeeded MKString.mkmodel
> succeeded MKDateTime.mkmodel
> *** FAILED *** MKEnums.mkmodel
Like Postgres, MS SQL Server has no concept of enums--so I added explicit=
support for them. My point is that you might be able to pull such support=
from MK's MS SQL Server code to use in MK's Postrges code.