Richard Broersma Jr wrote:
>> It seems almost impossible for the driver to solve the problems perfectly.
>> Essentially it's a problem of the rule system.
>
> Yes, I was able to reproduce these bad results from within psql. I take it that this is a well
> know limitation of the rule system or should I try reporting it with my simple test case?
I've had little experience with the PostgreSQL rule system. Someone with much experience may
have the solution. Please report the problem to the other MLs.
<snip>
> My next question, is it possible to create an updatable view using triggers on a instead of rules?
AFAIK it's impossible though I'm not sure.
> postgres=# select * from vwife;
> id | name | dresssize
> ----+---------+-----------
> 3 | dodie | 13
> 4 | heather | 10
> 2 | katie | 11
> (3 rows)
>
> postgres=# update vwife
> set name = 'Katheryn',
> dresssize = 12
> where (id,name,dresssize)=(2,'katie',11);
> UPDATE 0
>
> postgres=# select * from vwife;
> id | name | dresssize
> ----+----------+-----------
> 3 | dodie | 13
> 4 | heather | 10
> 2 | Katheryn | 11
> ^^^^^^^^ <-- update 0 is false
> (3 rows)
>
> Partial updates is a very bad thing.
Yes it's a disaster.
regards,
Hiroshi Inoue