Re: list of extended statistics on psql

Hi Tomas,

Thanks for your comments and also revising patches.

On 2020/11/16 3:22, Tomas Vondra wrote:
> It's better to always post the whole patch series, so that cfbot can
> test it properly. Sending just 0003 separately kind breaks that.

I now understand how "cfbot" works so that I'll take care of that
when I send patches. Thanks.


> Also, 0003 seems to only tweak the .sql file, not the expected output,
> and there actually seems to be two places that mistakenly use \dx (so
> listing extensions) instead of \dX. I've fixed both issues in the
> attached patches.

Oops, sorry about that.

  
> However, I think the 0002 tests are better/sufficient - I prefer to keep
> it compact, not interleaving with the tests testing various other stuff.
> So I don't intend to commit 0003, unless there's something that I don't
> see for some reason.

I Agreed. 0002 is easy to modify test cases and check results than 0003.
Therefore, I'll go with 0002.

  
> The one remaining thing I'm not sure about is naming of the columns with
> size of statistics - N_size, D_size and M_size does not seem very clear.
> Any clearer naming will however make the tables wider, though :-/

Yeah, I think so too, but I couldn't get an idea of a suitable name for
the columns when I created the patch.
I don't prefer a long name but I'll replace the name with it to be clearer.
For example, s/N_size/Ndistinct_size/.

Please find attached patcheds:
   - 0001: Replace column names
   - 0002: Recreate regression test based on 0001


Regards,
Tatsuro Yamada