Nikhil Sontakke wrote:
There is however one thing that needs some attention. Both patches add
distinction between VIEW and TABLE objects for acls into parser and they
both do it differently. GRANT ON ALL works by adding ACL_OBJECT_VIEW and
tracks that object type in code (that was my original method in both
patches) while DefaultACLs uses method suggested by Stephen Frost which is
creating new enum with relation, view, function and sequence members (those
are object types for which both DefaultACLs and GRANT ON ALL are
applicable). The second method has advantage of minimal changes to existing
code.
I briefly looked at the DefaultACLs patch. Can you not re-use the
GrantStmt structure for the defaults purpose too? You might have to
introduce an "is_default" boolean similar to the "is_schema" boolean
that you have added in the "GRANT ON ALL" patch. If you think you can
re-use the GrantStmt structure, then we might as well stick with the
existing object type code and not add the enums in the DefaultACLs
patch too..
No we can't use the GrantStmt and I wasn't talking about using it. I was
talking about the change in GrantObjectType and differentiating VIEW and
TABLE in some code inside aclchk.c which people didn't like. We can use
the changed GrantObjectType in DefaultACLs and filter inapplicable types
inside C code as I do in GRANT ON ALL patch and it's what I did
originally, but submitted version of DefaultACLs behaves differently.
--
Regards
Petr Jelinek (PJMODOS)
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
