David Steele <david@pgmasters.net> writes:
> My honest, albeit novice, opinion is that it was a mistake to pull
> pg_audit from contrib. I know more than anyone that it had flaws,
> mostly owing to its implementation as an extension, but it also provided
> capability that simply does not exist right now. Recent conversations
> about PGXN demonstrate why that is not (currently) a good alternative
> for distributing extensions. That means pg_audit will have a more
> limited audience than it could have had. That's a shame, because people
> are interested in pg_audit, warts and all.
FWIW, I did not think that the consensus was that pg_audit could never
be in contrib. As I understood it, people felt that (1) the API might
not be sufficiently stable yet, and (2) there might be parts that should
be in core eventually. (2) is problematic only because we do not have
very good tools for migrating things from contrib to core without creating
user-visible compatibility issues; which is not pg_audit's fault but it's
still a constraint we have to keep in mind. So I'd encourage you to keep
working on it and trying to address the issues that were brought up.
regards, tom lane