On Wed, 6 May 2020 at 17:33, Tom Lane <tgl@xxxxxxxxxxxxx> wrote: > > Thom Brown <thom@xxxxxxxxx> writes: > > I guess I should read the docs more carefully. Shouldn't this be > > insignificant for a numeric value? > > That would require the range code to know whether the subtype considers > whitespace significant (or perhaps more usefully, whether an all-spaces > input is valid). We've stayed away from requiring range_in to have any > type-specific knowledge of that sort. > > Still, you could argue that the rule ought to be "an empty or all-blank > value must be quoted to distinguish it from an omitted bound" rather than > "an empty value must be quoted to distinguish it from an omitted bound". > > I'm not sure if we could get away with redefining that at this point, > though. It looks like range_out quotes such values already, so maybe a > change wouldn't be totally catastrophic (in the sense of breaking dump > files). But I still suspect there would be more people unhappy than > happy. Okay, I see that this isn't really worth changing. It's surprising behaviour, but I can see it's not a huge issue, and can be worked around anyway. Thanks -- Thom
