Re: Implementation of GROUPING SETS (T431: Extended grouping capabilities)

2009/9/3 Pavel Stehule <pavel(dot)stehule(at)gmail(dot)com>:
> 2009/9/3 Joshua Tolley <eggyknap(at)gmail(dot)com>:
>> On Thu, Sep 03, 2009 at 01:19:25AM +0400, Олег Царев wrote:
>>> After week-lengthed investigation, now i 'm sure - my level of
>>> qualification not enough for implementation task "GROUPING SETS".
>>> I require documentation about the executor and the planner, i can't
>>> understand scheme of work by source code.
>>> Many code, many cases, but very little information "what is it" and
>>> "how thos work". May be i stupid.
>>
>> I doubt you're stupid; a stupid person wouldn't know what GROUPING SETS meant,
>> wouldn't bother finding out, and certainly wouldn't bother trying to implement
>> it. It's very helpful that you've let us know you're not working on it. That
>> way Pavel, if he finds he has time and interest, or someone else, can work on
>> it without fear of conflicting with what you're doing. Thanks for your work;
>> please don't get discouraged!
>
> There some ways, how implement GROUPING SETS. Currently I don't would
> to continue on this topic without some sponsoring. It got too my time
> - and I am able to implement it only as some special variant of CTE.
> Second way is total grouping planner refactoring - what is out of me.

There supposed to be another way to implement, that is a "hybrid" way
of CTE approach and hash tables approach. As you noticed, in group
mode the planner may be untouchable which CTE helps to avoid and in
hash mode you've almost done it. I wouldn't like to agree to CTE based
approach totally, but the hybrid might be reasonable.

Regards,

>
> regard
> Pavel
>
>>
>> --
>> Joshua Tolley / eggyknap
>> End Point Corporation
>> http://www.endpoint.com
>>
>> -----BEGIN PGP SIGNATURE-----
>> Version: GnuPG v1.4.9 (GNU/Linux)
>>
>> iEYEARECAAYFAkqfM2QACgkQRiRfCGf1UMOHvgCgpzV9cvjhCWhzcmvRDbXjdBQ1
>> 4RYAn2E+ZLRLdho+c+ZFleslPrbyxsZN
>> =66vh
>> -----END PGP SIGNATURE-----
>>
>>
>

--
Hitoshi Harada