>> The behavior is not actually changed, but I had to move fillfactor away
>> because it cannot be declared on partitioned tables, it must be declared
>> on partitions only. Once there is a function to handle that it is pretty
>> easy to add the test.
>>
>> I can remove it but franckly there are only benefits: the default is now
>> tested by pgbench, the create query is smaller, and it would work with
>> older versions of pg, which does not matter but is good on principle.
>
> I am not saying that it is a bad check on its own, rather it might be
> good, but let's not do any unrelated change as that will delay the
> main patch. Once, we are done with the main patch, you can propose
> these as improvements.
I would not bother to create a patch for so small an improvement. This
makes sense in passing because the created function makes it very easy,
but otherwise I'll just drop it.
>> The user could do a -i with a version of pgbench and bench with another
>> one. I do that often while developing…
>
> I am not following what you want to say here especially ("pgbench and
> bench with another one"). Can you explain with some example?
While developing, I often run pgbench under development client against an
already created set of tables on an already created cluster, and usually
the server side on my laptop is the last major release from pgdg (ie 11.5)
while the pgbench I'm testing is from sources (ie 12dev). If I type
"pgbench" I run 11.5, and in the sources "./pgbench" runs the dev version.
--
Fabien.