Re: IMPORT FOREIGN SCHEMA statement

* Ronan Dunklau (ronan(dot)dunklau(at)dalibo(dot)com) wrote:
> The problem with checking if the type is the same is deciding where to stop.
> For composite types, sure it should be easy. But what about built-in types ?

Haven't we already decided to trust these based on server version
information? Look at what quals we find acceptable to push down to the
remote side... Certainly, composites built off of built-in types should
work.

> Or types provided by an extension / a native library ? These could theorically
> change from one release to another.

This is definitely an issue which we really need to figure out how to
address. I don't have any great ideas off-hand, but it feels like we'll
need a catalog and appropriate SQL-fu to allow users and extensions to
add other types and functions which can be pushed down.. For example,
it'd be great if PostGIS queries could be pushed down to the remote
server- and have that set up by the postgis extension on installation
(perhaps via a call-back hook which gives the extension a handle to the
remote server where it could interrogate the server and then a way to
store the information about what is allowed to be pushed to the remote
side, and how?).

That's certainly a rather complicated bit to address and I don't think
we should hold this up for that- but let's definitely be thinking about
how to add these things later and try to avoid putting anything in place
which would cause problems for us later...

Will try to come back to the rest of your questions later.. :)

Thanks,

Stephen