Re: Faster methods for getting SPI results

On 28 December 2016 at 12:32, Jim Nasby <Jim(dot)Nasby(at)bluetreble(dot)com> wrote:
> On 12/27/16 9:10 PM, Craig Ringer wrote:
>>
>> On 28 December 2016 at 09:58, Jim Nasby <Jim(dot)Nasby(at)bluetreble(dot)com> wrote:
>>
>>> I've looked at this some more, and ITSM that the only way to do this
>>> without
>>> some major surgery is to create a new type of Destination specifically
>>> for
>>> SPI that allows for the execution of an arbitrary C function for each
>>> tuple
>>> to be sent.
>>
>>
>> That sounds a lot more sensible than the prior proposals. Callback driven.
>
>
> Are there other places this would be useful? I'm reluctant to write all of
> this just to discover it doesn't help performance at all, but if it's useful
> on it's own I can just submit it as a stand-alone patch.

I don't have a use for it personally. In BDR and pglogical anything
that does work with nontrivial numbers of tuples uses lower level
scans anyway.

I expect anything that uses the SPI to run arbitrary user queries
could have a use for something like this though. Any PL, for one.

--
Craig Ringer http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services