On 11/02/2011 09:56 AM, Tom Lane wrote:
> Andrew Dunstan<andrew@dunslane.net> writes:
>> On 11/02/2011 03:16 AM, Valentine Gogichashvili wrote:
>>> Putting aside arguments like "it is not a good idea to use * because
>>> it generates not sustainable code especially in case when you extend
>>> table structure", I think this construct would be really nice for
>>> building ROWs, for example in plpgsql triggers or in conditions for
>>> big update statements:
>>>
>>> IF (NEW.* EXCLUDING ( last_modified ) ) IS DISTINCT FROM (OLD.*
>>> EXCLUDING ( last_modified ) ) THEN NEW.last_modified =
>>> clock_timestamp() ; END IF
>> That's a very good use case. I could certainly have used this in the past.
> Well ... this is inventing use cases that have nothing to do with the
> proposed feature and are entirely incapable of being supported by the
> proposed implementation. And I'm not sure why we should put aside the
> argument that this is only a good idea in ad-hoc queries, either.
>
>
Well, yes, you're right that it's not covered by the original feature. I
guess I got interested because a couple of years ago I had to write some
triggers in PLPerl and with much lower efficiency to achieve the same
effect as this.
cheers
andrew