Re: JSON for PG 9.2

On tis, 2011-12-13 at 00:06 -0800, Peter van Hardenberg wrote:
> On Mon, Dec 12, 2011 at 9:25 PM, Peter Eisentraut <peter_e(at)gmx(dot)net> wrote:
> > On mån, 2011-12-12 at 16:51 -0800, Peter van Hardenberg wrote:
> > You don't need a new PL to do that. The existing PLs can also parse
> > JSON. So that's not nearly enough of a reason to consider adding this
> > new PL.
>
> PL/V8 is interesting because it is very fast, sandboxed, and well
> embedded with little overhead.
>
> My experience with PL/Python and PL/Perl has not been thus, and
> although they are handy if you want to break out and run system work,
> they're not the kind of thing I'd consider for defining performant
> operators with.

Some performance numbers comparing a valid_json() functions implemented
in different ways would clarify this. I wouldn't be surprised if PL/V8
won, but we need to work with some facts.