On 2015-07-16 PM 12:43, Tom Lane wrote:
>
> The basic issue here is "how can a user control which functions/operators
> can be sent for remote execution?". While it's certainly true that
> sometimes you might want function-by-function control of that, Paul's
> point was that extension-level granularity would be extremely convenient
> for PostGIS, and probably for other extensions.
Perhaps just paranoid but is the extension version number any significant?
I guess that if a function name is matched locally and declared safe to
send but doesn't really exist on the remote end or does exist but has
different behavior, then that can't be expected to work or work correctly.
But it seems difficult to incorporate the version number into chosen
approach of matching functions anyway.
Thanks,
Amit