On Sat, 2013-12-07 at 12:27 -0500, Stephen Frost wrote:
> * Jeff Davis (pgsql@j-davis.com) wrote:
> > The behavior of an extension should not depend on how it was installed.
> >
> > The kind of "extension" being described by Stephen will:
> >
> > * Not be updatable by doing "ALTER EXTENSION foo UPDATE TO '2.0'"
>
> ... [ reason ] ...
> > * Dump out objects that wouldn't be dumped if they had installed the
> > extension using the filesystem
> ... [ reason ] ...
I understand there are reasons, but I'm having a hard time getting past
the idea that "I have extension foo v1.2" now needs to be qualified with
"installed using SQL" or "installed using the filesystem" to know what
you actually have and how it will behave.
Stepping back, maybe we need to do some more research on existing
SQL-only extensions. We might be over-thinking this. How many extensions
are really just a collection of functions on existing types? If you
define a new data type, almost all of the functions seem to revolve
around C code -- not just to define the basic data type, but also the
GiST support routines, which then mean you're implementing operators in
C too, etc.
Perhaps we should first focus on making SQL-only extensions more useful?
Regards,Jeff Davis