Re: VACUUM fails to parse 0 and 1 as boolean value

Hi,

On 2019-05-21 16:00:25 +0900, Kyotaro HORIGUCHI wrote:
> At Tue, 21 May 2019 14:31:32 +0900, Michael Paquier <michael(at)paquier(dot)xyz> wrote in <20190521053132(dot)GG1921(at)paquier(dot)xyz>
> > On Mon, May 20, 2019 at 09:55:59AM -0400, Robert Haas wrote:
> > > Well, it's confusing that we're not consistent about which spellings
> > > are accepted. The GUC system accepts true/false, on/off, and 0/1, so
> > > it seems reasonable to me to standardize on that treatment across the
> > > board. That's not necessarily something we have to do for v12, but
> > > longer-term, consistency is of value.
> >
> > +1.
> >
> > Note: boolean GUCs accept a bit more: yes, no, tr, fa, and their upper
> > case flavors, etc. These are everything parse_bool():bool.c accepts
> > as valid values.
>
> Yeah, I agree for longer-term. The opinion was short-term
> consideration on v12. We would be able to achieve full
> unification on sub-applications in v13 in that direction. (But I
> don't think it's good that apps pass-through options then server
> checkes them..)

To me it is odd to introduce an option, just to revamp the accepted
style of arguments in the next release. I think we ought to just clean
this up now.

Greetings,

Andres Freund