Re: More on view updates and inverse views
From: Vadim Tropashko <vadimtro_at_gmail.com>
Date: Tue, 8 Sep 2009 12:16:38 -0700 (PDT)
Message-ID: <dde2aed2-8d1f-4793-b481-a435383934e4_at_m7g2000prd.googlegroups.com>
Date: Tue, 8 Sep 2009 12:16:38 -0700 (PDT)
Message-ID: <dde2aed2-8d1f-4793-b481-a435383934e4_at_m7g2000prd.googlegroups.com>
On Sep 8, 9:35 am, "Walter Mitty" <wami..._at_verizon.net> wrote:
> This makes the projection view
> "invertable" even though without the default value it would not be
> invertable.
And what is default value in relational terms? Suppose we have relation x over attributes p and q, and the q attribute is "default". Then, we essentially have a union of two(!) base relations:
- A restriction of the original binary relation x, where no tuple contains default value with
- y ^ [q] default, where y is some unary relation with attribute p.
In other words we have a view
x v (y ^ [q] default)
Then, I'm not quite convinced that the view
(x v (y ^ [q] default)) v [p]
is updatable. The "default" is just some value, which constraints would specifially instruct the relational system that it behaves like default? Received on Tue Sep 08 2009 - 21:16:38 CEST