Thread: Post to hacker list

Post to hacker list

From
Francisco Reyes
Date:
Sent the email below to the hackers list. Since I was not in the list it
gave me a message saying it needed to be authorized.

I then tried to join the hackers list in case people answered and did not
CC me personally. Never got the confirmation email from the list. Anything
special about the list? Is it a closed list?


I am basically trying to see what it would take in $$ to get someone to
improve inherittance. If the actual work would be too expensive maybe
could at least pay someone for a study of what exactly needs to be done
and then maybe leave the actual work for another phase or see if someone
would be willing to do it for a fee I could afford.

 ---------- Forwarded message ----------
Date: Thu, 18 Mar 2004 10:26:25 +0000 (GMT)
From: Francisco Reyes <francisco@natserv.net>
To: pgsql-hackers@postgresql.org
Subject: Inherited tables

On a recent thread Stephan Szabo mentioned some issues with inheritance.

***********
> On Thu, 19 Feb 2004, Stephan Szabo wrote:
>
> > There are two separate things here that are gotchas
> > The first is that unique constraints don't inherit, and foreign keys must
> > refer to a unique constraint.
>
> > These are both deficiencies in inheritance and the constraints in
> > question.
***********

Anyone could comment on those problems?
I am working on a design which will heavily use inherittance and I think
on the long run the limitations above could make things to be more work
for me.

I would be willing to put some money forward to have someone work on
improving those limitations. Anyone knows who may be a good candidate to
work on this?

Re: Post to hacker list

From
Tino Wildenhain
Date:
Hi Francisco,

Francisco Reyes wrote:
> Sent the email below to the hackers list. Since I was not in the list it
> gave me a message saying it needed to be authorized.
>
> I then tried to join the hackers list in case people answered and did not
> CC me personally. Never got the confirmation email from the list. Anything
> special about the list? Is it a closed list?
>
>
> I am basically trying to see what it would take in $$ to get someone to
> improve inherittance. If the actual work would be too expensive maybe
> could at least pay someone for a study of what exactly needs to be done
> and then maybe leave the actual work for another phase or see if someone
> would be willing to do it for a fee I could afford.
>
This would be great!
Maybe we could get some donations if only we know what needs to be done.

Hope soneone will answer.

Regards
Tino Wildenhain


Re: Post to hacker list

From
Francisco Reyes
Date:
On Sat, 20 Mar 2004, Tino Wildenhain wrote:

> This would be great!
> Maybe we could get some donations if only we know what needs to be done.
> Hope soneone will answer.

I have continued researching for both what ist he problem and who could
work on it.

Of the people I have exchanged emails with Stephan Szabo seems to have a
decent understanding of the internals of the problem and whatever little I have
learnt about the problem and the magnitude of possible solutions came from
him emails.

I like the idea of doing some type of donation system for those of us
interested in getting inheritance improved. What I have seen so far makes
inheritance a risky proposition to use.

For example if you have
parent table
common field1 primary key
common field2

inherited table
unique field1
unique field2

It is NOT possible to have duplicates in common field1 if you insert into
the parent table, but it IS possible to create duplicaes in common field 1
if you do the insert with repeated fields in inherited table.

In a design I am working on I would use a serial for that column, which
works accros parent/inherited tables. But the risk is that it is possible
that if I EVER make a mistake in my coding and write to the primary key
column instead of getting the value from the sequence the database won't
catch the problem.

For a system where only a handfull of programmers are working it may not
be terribly bad, but if you had a system where teams of people are working
having the database not be able to inforce the uniqueness accross
parent/inheritted table is a problem.