RE: Revisited: Transactions, insert unique.

> -----Original Message-----
> From: pgsql-general-owner@hub.org [mailto:pgsql-general-owner@hub.org]On
> Behalf Of Ross J. Reedstrom
>
> And this interpretation will guide the developers in _extending_
> the standard in a consistent way. I know, because the developers that
> implemented the constraints for 7.0 used this (and the SQL3 spec) as
> guides. How's that?
>

I don't know what is standard.
However as far as I see,few people prefer entire rollback on abort.
The problem is that PostgreSQL lacks a per statement rollback
functionality and unfortunately it isn't easy to implement.
Vadim has already planned the implementation. AFAIK one of the
purpose of WAL is to implement savepoint functionality. Savepoint
functionality would enable per statement rollback functionality easily.

The following is an extract of Vadim's posting about 10 months ago.

Re: [HACKERS] RE: [GENERAL] Transaction logging
Well, I'm thinking about WAL last two weeks. Hiroshi pointed me
problems in my approach to savepoints (when a tuple was marked
for update and updated after it) and solution would require
new tid field in header and both t_cmin/t_cmax => bigger header.
I don't like it and so I switched my mind -:).

Regards.

Hiroshi Inoue
Inoue@tpf.co.jp