Re: [COMMITTERS] pgsql: Basic documentation for ROLEs.

Alvaro Herrera <alvherre@alvh.no-ip.org> writes:
> Would you post your whole to-do list for roles?

This is more or less verbatim (now you know what kind of notes I keep):



Do we want ROLE to be GUC_REPORT?

Should RESET ALL reset ROLE??

Got some problems with rolling back SET SESSION AUTH (won't restore prior SET
ROLE if any) and with rolling back an xact in which both were set (order
dependence, though I think net effect is same as above).

createuser script needs enhancement?  dropuser?

do we need a bigger is_member cache?

shouldn't this work:
regression=# create role r;
CREATE ROLE
regression=# grant r to public;
ERROR:  role "public" does not exist
Not clear how this squares with rules against circular grants though;
don't we want to act as though public is granted to every role?

Should has_role(foo, PUBLIC) always succeed?  If so could probably eliminate a
couple of special cases in information_schema.

information_schema still needs a look, particularly role_column_grants

pg_dumpall support
Need ON DELETE pg_shadow rule for existing dump files??
There's a DELETE FROM pg_group too :-(

what is the grantor col in pg_auth_members really for?  Either we
don't need it or we're failing to follow semantics.  Shouldn't
revoking admin option result in tracing and revoking all grants
from that grantor?

DOCS ...
note samerole addition to pg_hba.conf
tutorial/syscat.source

regression tests?

Add new \d displays in psql?



> Also, what do we have in the open items list?  The current list at
> http://candle.pha.pa.us/cgi-bin/pgopenitems is outdated.

That is Bruce's bailiwick.  He told me yesterday that he was intending
to update it for 8.1 before leaving for OSCON.
        regards, tom lane