Re: [v9.2] Fix leaky-view problem, part 1

On Sat, Jul 2, 2011 at 3:46 PM, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> Robert Haas <robertmhaas(at)gmail(dot)com> writes:
>> On Sat, Jul 2, 2011 at 1:54 PM, Kohei KaiGai <kaigai(at)kaigai(dot)gr(dot)jp> wrote:
>>> BTW, regarding to the statement support for security barrier views,
>>> the following syntax might be more consistent with existing ones:
>>>  CREATE VIEW view_name WITH ( param [=value]) AS query ... ;
>>> rather than
>>>  CREATE SECURITY VIEW view_name AS query ...;
>>>
>>> Any comments?
>
>> I think I mildly prefer CREATE SECURITY VIEW to the parameter syntax
>> in this case, but I don't hate the other one.
>
> The WITH idea seems a bit more future-proof; in particular it would
> easily accommodate specifying a security type, if we decide we need
> various levels of leak-proof-ness.

Or other kinds of view options. I'm not going to argue against that
too forcefully, since I've advocated introducing that sort of syntax
elsewhere. I think it's mostly that I thought this feature might be
significant enough to merit a syntax that would make it a little more
prominent, but perhaps not.

--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company