Re: RETURNING syntax for COPY

On 5/8/13 12:33 PM, Dimitri Fontaine wrote:
> Karol Trzcionka <karlikt(at)gmail(dot)com> writes:
>> as a continuation of my proposal expanding RETURNING syntax by
>
> What about implementing support for OLD/NEW in per-statement triggers? I
> guess you would expose the data via a SRF.

Per statement NEW/OLD is an interesting case, in that it shares some of the same challenges; namely how to store the NEW and OLD recordsets efficiently. I've wondered if there'd be some way to do that by just storing a list of CTIDs (not sure if that'd work with HOT for OLD though).

I still like the idea of being able to exclude certain records during COPY though; not writing a tuple will always be more efficient than creating one and then nuking it after the fact. There's a similar argument to be made about in-line transforms too.
--
Jim C. Nasby, Data Architect jim(at)nasby(dot)net
512.569.9461 (cell) http://jim.nasby.net