Re: Extension support for postgres_fdw - Mailing list pgsql-hackers

From Heikki Linnakangas
Subject Re: Extension support for postgres_fdw
Date
Msg-id 55858C59.8020802@iki.fi
Whole thread Raw
In response to Extension support for postgres_fdw  (Paul Ramsey <pramsey@cleverelephant.ca>)
List pgsql-hackers
On 06/20/2015 10:20 AM, Paul Ramsey wrote:
> I would like to enhance the postgres_fdw to allow more complete support for user-defined types.
>
> Right now, postgres_fdw already does a good job of passing user-defined type data back and forth, which is pretty
nice.However, it will not pass functions or operators that use user-defined types to the remote host. For a extension
likePostGIS, that means that spatial filters cannot be executed on remote servers, which makes FDW not so useful for
PostGIS.
>
> I think the postgres_fdw extension should pass user-defined functions and operators, but only when it knows those
functionsand operators exist at the remote. One way would be to ask the remote what extensions it has, but the overhead
ofdoing that is a bit high. A simpler way would be to just have the DBA declare what extensions the remote will have,
whenshe creates the server definition, for example:
 
>
> CREATE SERVER fire_department_server
>    FOREIGN DATA WRAPPER postgres_fdw
>    OPTIONS (
>      host       'fire.city.gov',
>      dbname     'infrastructure',
>      port       '5432',
>      extensions 'postgis, seg'
> );
>
> Once the local host knows what extensions to expect on the remote side, it can retain functions and operators in
thoseextensions in the set of remote restrictions and deparse them for use in the query of the remote. Basically,
everywherethere is a call to is_builtin(Oid oid) now, there's also be a call to is_allowed_extension() (or somesuch) as
well.

The SQL/MED specification has a concept of routine mappings, for mapping 
functions. I'm not sure if it'd be better to implement CREATE ROUTINE 
MAPPING, and adapt it to operators too, or invent something entirely new 
and PostgreSQL-specific. But CREATE ROUTINE MAPPING at least deserves a 
look.
- Heikki




pgsql-hackers by date:

Previous
From: Tomas Vondra
Date:
Subject: Re: pretty bad n_distinct estimate, causing HashAgg OOM on TPC-H
Next
From: Tom Lane
Date:
Subject: Re: error message diff with Perl 5.22.0