Re: CustomScan in a larger structure (RE: CustomScan support on readfuncs.c)

> -----Original Message-----
> From: pgsql-hackers-owner(at)postgresql(dot)org
> [mailto:pgsql-hackers-owner(at)postgresql(dot)org] On Behalf Of Robert Haas
> Sent: Thursday, February 04, 2016 11:39 AM
> To: Kaigai Kouhei(海外 浩平)
> Cc: Andres Freund; Amit Kapila; pgsql-hackers
> Subject: Re: CustomScan in a larger structure (RE: [HACKERS] CustomScan support
> on readfuncs.c)
>
> On Wed, Feb 3, 2016 at 8:00 PM, Kouhei Kaigai <kaigai(at)ak(dot)jp(dot)nec(dot)com> wrote:
> >> Well, looking at this a bit more, it seems like the documentation
> >> you've written here is really misplaced. The patch is introducing a
> >> new facility that applies to both CustomScan and ForeignScan, but the
> >> documentation is only to do with CustomScan. I think we need a whole
> >> new chapter on extensible nodes, or something. I'm actually not
> >> really keen on the fact that we keep adding SGML documentation for
> >> this stuff; it seems like it belongs in a README in the source tree.
> >> We don't explain nodes in general, but now we're going to have to try
> >> to explain extensible nodes. How's that going to work?
> >>
> > The detail of these callbacks are not for end-users, administrators and
> > so on except for core/extension developers. So, I loves idea not to have
> > such a detailed description in SGML.
> > How about an idea to have more detailed source code comments close to
> > the definition of ExtensibleNodeMethods?
> > I haven't seen the src/backend/nodes/README yet, and it has only 6 updates
> > history from Postgres95 era. I guess people may forget to update README
> > file if description is separately located from the implementation.
>
> Hmm, that might work, although that file is so old that it may be
> difficult to add to. Another idea is: maybe we could have a header
> file for the extensible node stuff and just give it a really long
> header comment.
>
At this moment, I tried to write up description at nodes/nodes.h.
The amount of description is about 100lines. It is on a borderline
whether we split off this chunk into another header file, in my sense.

On the other hands, I noticed that we cannot omit checks for individual
callbacks on CustomXXXX node type, ExtensibleNodeMethods is embedded in
the CustomXXXXMethods structure, thus we may have CustomXXXX node with
no extensible feature.
This manner is beneficial because extension does not need to register
the library and symbol name for serialization. So, CustomScan related
code still checks existence of individual callbacks.

Thanks,
--
NEC Business Creation Division / PG-Strom Project
KaiGai Kohei <kaigai(at)ak(dot)jp(dot)nec(dot)com>

> --
> Robert Haas
> EnterpriseDB: http://www.enterprisedb.com
> The Enterprise PostgreSQL Company
>
>
> --
> Sent via pgsql-hackers mailing list (pgsql-hackers(at)postgresql(dot)org)
> To make changes to your subscription:
> http://www.postgresql.org/mailpref/pgsql-hackers