On Thu, Sep 29, 2011 at 4:22 AM, Peter Eisentraut <peter_e@gmx.net> wrote:
>> Not sure what to do about this. =A0Is it okay to suppose that collation
>> can be ignored when matching to a collation-less index?
>
> That sounds correct on first reading.
>
Doesn't this depend on the semantics of the ? operator?
Hypothetically if there was an operator like ?< which returned a list
of hstore keys that were < the argument then ?< would indeed depend on
the collation used even if hstore didn't do collations. If there was
an index type on hstore which could handle ?< then it would need to
have the right collation to be usable.
Of course we know ? doesn't depend on the collation but where is that
information? I suspect we should actually have an explicit flag for
each operator.
--=20
greg