Re: Parallel Aggregates for string_agg and array_agg

On 03/29/2018 03:09 PM, David Rowley wrote:
> On 30 March 2018 at 02:00, Tomas Vondra <tomas.vondra@2ndquadrant.com> wrote:
>> On 03/29/2018 05:49 AM, David Rowley wrote:
>>> Attached is v9, which is based on Tom's v8 but includes the new tests
>>> and I think the required fix to disable use of the serial/deserial
>>> function for array_agg().
>>>
>>
>> I have only looked at the diff, but it seems fine to me (in the sense
>> that it's doing the right checks to disable parallelism only when really
>> needed).
>>
>> FWIW I wonder if we might simply fallback to input/output functions when
>> the send/receive functions are not available. Not sure it's worth it.
> 
> I think it's a corner case to have a type without send/receive, but I
> might just be lacking imagination.
> 

Not sure, but it's a valid case. OTOH we probably don't need to obsess
about handling it perfectly.

> I meant to mention earlier that I coded
> agg_args_have_sendreceive_funcs() to only check for send/receive
> functions. Really we could allow a byval types without send/receive
> functions, since the serial/deserial just send the raw datums in that
> case, but then the function becomes
> agg_byref_args_have_sendreceive_funcs(), which seemed a bit obscure,
> so I didn't do that.  Maybe I should?
> 

I'd do that. Not sure the function name needs to change, but perhaps
agg_args_support_sendreceive() would be better - it covers both byref
types (which require send/receive functions) and byval (which don't).

regards

-- 
Tomas Vondra                  http://www.2ndQuadrant.com
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services