Re: Finer Extension dependencies

"Kevin Grittner" <Kevin(dot)Grittner(at)wicourts(dot)gov> writes:
> I gather from previous posts that the intent isn't to allow different
> packages from different authors to provide a common and compatible
> feature; but what happens in the current design if someone
> accidentally or maliciously produces an extension which provides the
> same feature name as another extension?

It's not about that, it's more like the features/require/provide
concepts in Lisp, except that we're not using them to load files.

The goal really is to avoid a feature matrix and a version policy with
comparators, yet be able to depend on features that got implemented
after the first release of an extension.

> Would we need some registry?

That being said, we still have a single namespace for extensions and
their features, so a registry would help, yes.

--
Dimitri Fontaine
http://2ndQuadrant.fr PostgreSQL : Expertise, Formation et Support