On Wed, Sep 19, 2012 at 2:17 PM, Noah Misch <noah@leadboat.com> wrote:
> Sounds fine for now. I suspect the better change would be to make
> AcceptInvalidationMessages() unconditional in LockRelationOid() and friends.
> There's no reason to desire recent ACLs only when opening by name.
I agree, on both counts. I think I failed to realize when doing that
refactoring that I was reducing the number of cases where
AcceptInvalidationMessages() would actually be called. AFAICS, the
only reason why we have such complicated rules for calling that
function is that it's traditionally been expensive. But that should
be much less true due now due to improvements in 9.2 (cf commit
b4fbe392f8ff6ff1a66b488eb7197eef9e1770a4). So we can probably afford
to be a little less byzantine about the way we do this now.
--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company