Re: Feature suggestions (long)

From: Martijn van Oosterhout <kleptog(at)svana(dot)org>
To: "Jim C(dot) Nasby" <jim(at)nasby(dot)net>
Cc: Zeugswetter Andreas SB SD <ZeugswetterA(at)spardat(dot)at>, pgsql-hackers(at)postgresql(dot)org
Subject: Re: Feature suggestions (long)
Date: 2003-05-20 15:23:44
Message-ID: 20030520152344.GH4069@svana.org
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

On Tue, May 20, 2003 at 10:02:24AM -0500, Jim C. Nasby wrote:
> On Tue, May 20, 2003 at 12:40:00AM +1000, Martijn van Oosterhout wrote:
> > Anyway, the general trend seems to be against the idea so I may as well go
> > think of something else :)
>
> I'm disappointed to hear that. Having no way to effectively partition
> data is a real pain in pgsql, and your proposal would adress that. Yes,
> you can build it yourself by creating the view and all the rules by
> hand, but that has a lot of drawbacks:

I agree, there is a lot of potential here. And I don't beleive it would be
too much work as most of the infrastructure is already there. At this stage
I'm just wondering if it will go on the TODO list. I propose that the
following items be added:

* Improve the planner to take CHECK constraints into account to prune the plan.
* Allow a single index to index multiple tables (also for inherited PRIMARY KEYS)
* Allow partitioning of table into multiple subtables

The first two items would be useful in their own right. With them the final
one would be straight forward. I'd be prepared to put some effort into this
if there is some indication it would be accepted.

> I don't know what the policies for patches are, but I'd hope that the
> core team would consider adding this functionality, especially since a
> first-round implimentation can be done entirely with rules (or so it
> seems).

Well, I think the policy is 'if you write the code you have a better chance
to have it accepted' :) So, if it's likely to be accepted then we only need
to find someone to code it. Given the other priorities currently I think
waiting for the core team to write it would be futile (unless you can
convince someone like IBM to give the core team money to write it).

Right now I'd be happy if the anonymous CVS server would talk to me :)

By the way, has anyone given thought to user-defined storage managers? Apart
from allowing backward compatable table access, you could implement a simple
version of partitioning that doesn't take advantage of planner tricks.

Have a nice day,
--
Martijn van Oosterhout <kleptog(at)svana(dot)org> http://svana.org/kleptog/
> "the West won the world not by the superiority of its ideas or values or
> religion but rather by its superiority in applying organized violence.
> Westerners often forget this fact, non-Westerners never do."
> - Samuel P. Huntington

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Jan Wieck 2003-05-20 15:25:22 Re: Calling external program from trigger
Previous Message Jim C. Nasby 2003-05-20 15:07:17 Re: Interessting problem