> On Apr 23, 2020, at 8:48 AM, Adrian Klaver wrote:
>
> No.
Ah, thanks a lot for that info.
It's not that I really normally want to have bazillions of params.
But we have some moderately large inserts, that are showing a sudden non-linear
dropoff when scaling up, and a suspicion that the perf
On Thursday, April 23, 2020, Adrian Klaver
wrote:
> On 4/23/20 7:33 AM, Scott Ribe wrote:
>
>> In libpq, PQexecParams has nParams as type int. So on any reasonable
>> platform, that's at least 4 bytes. My question then is: when I see
>> documented limits of 65535 params in various drivers and lib
On 4/23/20 7:33 AM, Scott Ribe wrote:
In libpq, PQexecParams has nParams as type int. So on any reasonable platform,
that's at least 4 bytes. My question then is: when I see documented limits of
65535 params in various drivers and libraries, that is NOT a restriction of
libpq nor of the protoc
In libpq, PQexecParams has nParams as type int. So on any reasonable platform,
that's at least 4 bytes. My question then is: when I see documented limits of
65535 params in various drivers and libraries, that is NOT a restriction of
libpq nor of the protocol, but rather an arbitrary limit of the
