On Mon, Mar 16, 2020 at 11:07:37AM +0900, Atsushi Torikoshi wrote:
> As far as I read the reloptions.c, autovacuum_vacuum_cost_delay,
> autovacuum_vacuum_scale_factor and autovacuum_analyze_scale_factor
> are the members of relopt_real, so their type seems the same, real.
In this case, the parsing uses parse_real(), which is exactly the same
code path as what real GUCs use.
> But the manual about storage parameters[1] says two of their type
> are float4 and the other is floating point.
>
> I think using float4 on storage parameters[1] are not consistent
> so far, how about changing these parameters type from float4 to
> floating point if there are no specific reasons using float4?
That's a good idea, so I am fine to apply your patch as float4 is a
data type. However, let's see first if others have more comments or
objections.
--
Michael