Re: alter user/role CURRENT_USER

On Fri, Oct 24, 2014 at 11:29 AM, Kyotaro HORIGUCHI
<horiguchi(dot)kyotaro(at)lab(dot)ntt(dot)co(dot)jp> wrote:
> - 0001-ALTER-ROLE-CURRENT_USER_v2.patch - the patch.

+RoleId: CURRENT_USER { $$ = "current_user";}
+ | USER { $$ = "current_user";}
+ | CURRENT_ROLE { $$ = "current_user";}
+ | SESSION_USER { $$ = "session_user";}

This is kind of ugly, and it means you can't distinguish between a
CURRENT_USER keyword and a quoted user name "current_user". It's a
legitimate user name, so the behavior of the following now changes:

CREATE ROLE "current_user";
ALTER ROLE "current_user" SET work_mem='10MB';

There ought to be a better way to represent this than using magic string values.

----
> It accepts CURRENT_USER/USER/CURRENT_ROLE/SESSION_USER.

On a stylistic note, do we really want to support the alternative
spellings of CURRENT_USER, like CURRENT_ROLE and USER? The SQL
standard is well-hated for inventing more keywords than necessary.
There is no precedent for using/allowing these aliases in PostgreSQL
DDL commands, and ALTER USER & ROLE aren't SQL standard anyway. So
maybe we should stick with just accepting one canonical syntax
instead.

I think the word USER may reasonably arise from an editing mistake,
ALTER USER USER does not seem like sane syntax that should be
accepted.

----
From your original email:

> - Modify syntax of ALTER USER so as to be an alias of ALTER ROLE.
> - Added ALL syntax as user name to ALTER USER.

But should ALTER USER ALL and ALTER ROLE ALL really do the same thing?
A user is a role with the LOGIN option. Every user is a role, but not
every role is a user. I suspect that ambiguity was why ALTER USER ALL
wasn't originally implemented.

Regards,
Marti