Hi Alvaro,
On Fri, Mar 29, 2024 at 2:04 AM Alvaro Herrera <alvherre@alvh.no-ip.org> wrote:
> On 2024-Mar-28, Amit Langote wrote:
>
> > Here's patch 1 for the time being that implements barebones
> > JSON_TABLE(), that is, without NESTED paths/columns and PLAN clause.
> > I've tried to shape the interfaces so that those features can be added
> > in future commits without significant rewrite of the code that
> > implements barebones JSON_TABLE() functionality. I'll know whether
> > that's really the case when I rebase the full patch over it.
>
> I think this barebones patch looks much closer to something that can be
> committed for pg17, given the current commitfest timeline. Maybe we
> should just slip NESTED and PLAN to pg18 to focus current efforts into
> getting the basic functionality in 17. When I looked at the JSON_TABLE
> patch last month, it appeared far too large to be reviewable in
> reasonable time. The fact that this split now exists gives me hope that
> we can get at least the first part of it.
Thanks for chiming in. I agree that 0001 looks more manageable.
> (A note that PLAN seems to correspond to separate features T824+T838, so
> leaving that one out would still let us claim T821 "Basic SQL/JSON query
> operators" ... however, the NESTED clause does not appear to be a
> separate SQL feature; in particular it does not appear to correspond to
> T827, though I may be reading the standard wrong. So if we don't have
> NESTED, apparently we could not claim to support T821.)
I've posted 0002 just now, which shows that adding just NESTED but not
PLAN might be feasible.
--
Thanks, Amit Langote