Re: Partial aggregates pushdown

Daniel Gustafsson писал 2021-11-15 13:16:
>> On 3 Nov 2021, at 15:50, Alexander Pyhalov <a(dot)pyhalov(at)postgrespro(dot)ru>
>> wrote:
>>
>> Daniel Gustafsson писал 2021-11-03 16:45:
>>>> On 2 Nov 2021, at 10:12, Alexander Pyhalov
>>>> <a(dot)pyhalov(at)postgrespro(dot)ru> wrote:
>>>> Updated and rebased patch.
>>> + state = (Int128AggState *) palloc0(sizeof(Int128AggState));
>>> + state->calcSumX2 = false;
>>> +
>>> + if (!PG_ARGISNULL(0))
>>> + {
>>> +#ifdef HAVE_INT128
>>> + do_int128_accum(state, (int128) PG_GETARG_INT64(0));
>>> +#else
>>> + do_numeric_accum(state, int64_to_numeric(PG_GETARG_INT64(0)));
>>> +#endif
>>> This fails on non-INT128 platforms as state cannot be cast to
>>> Int128AggState
>>> outside of HAVE_INT128; it's not defined there. This needs to be a
>>> PolyNumAggState no?
>>
>> Hi.
>> Thank you for noticing this. It's indeed fails with
>> pgac_cv__128bit_int=no.
>> Updated patch.
>
> The updated patch also fails to apply now, but on the catversion.h
> bump. To
> avoid having to rebase for that I recommend to skip that part in the
> patch and
> just mention the need in the thread, any committer picking this up for
> commit
> will know to bump the catversion so there is no use in risking
> unneccesary
> conflicts.

I've updated patch - removed catversion dump.

--
Best regards,
Alexander Pyhalov,
Postgres Professional