> On Thu, Feb 09, 2023 at 02:30:34PM +0100, Peter Eisentraut wrote:
> On 07.02.23 21:14, Sergei Kornilov wrote:
> > It seems a little strange to me that with const_merge_threshold = 1, such a test case gives the same result as with
const_merge_threshold= 2
>
> What is the point of making this a numeric setting? Either you want to
> merge all values or you don't want to merge any values.
At least in theory the definition of "too many constants" is different
for different use cases and I see allowing to configure it as a way of
reducing the level of surprise here. The main scenario for a numerical
setting would be to distinguish between normal usage with just a handful
of constants (and the user expecting to see them represented in pgss)
and some sort of outliers with thousands of constants in a query (e.g.
as a defence mechanism for the infrastructure working with those
metrics). But I agree that it's not clear how much value is in that.
Not having strong opinion about this I would be fine changing it to a
boolean option (with an actual limit hidden internally) if everyone
agrees it fits better.