On Wed, Jan 15, 2014 at 10:27 AM, Vik Fearing <vik(dot)fearing(at)dalibo(dot)com> wrote:
> On 12/26/2013 06:10 PM, David Fetter wrote:
>> On Tue, Dec 24, 2013 at 07:25:01PM +0100, Vik Fearing wrote:
>>> I was recently annoyed that I had to do
>>>
>>> CREATE TABLESPACE x LOCATION y;
>>> ALTER TABLESPACE x SET (random_page_cost = z);
>>>
>>> The attached patch is a quick n' dirty extension to allow the SET
>>> directly on the CREATE statement.
>>>
>>> CREATE TABLESPACE x LOCATION y SET (random_page_cost = z);
>> That should probably be WITH instead of SET for consistency with other
>> similar DDL.
>
> Here is version 2 of the patch, which uses WITH instead of SET, and also
> adds to the documentation.
I just had a quick look at this patch, no testing at all. I am seeing
that regression tests are still missing here, they should be added in
src/test/regress/input/tablespace.source. Then, for the use of either
WITH or SET... For example CREATE FUNCTION uses SET for a
configuration parameter, and ALTER FUNCTION is consistent with that.
So SET makes more sense to be consistent with CREATE TABLESPACE? This
patch should not be modified once again as long as there are no more
opinions though...
Regards,
--
Michael