Subject | Re: Nulls in CHECK Constraints |
---|---|
Author | paulruizendaal |
Post date | 2005-10-26T11:08:34Z |
I have had a look at the (FB15) code. Changing this behaviour to
standard compliance is a three-line fix in dsql/gen.cpp I would
assume the same holds for FB2 and Vulcan.
Now we have to think about our installed base and how me might
migrate CHECK behaviour without leaving our beloved users up the
proverbial creek without tools.
Paul
--- In Firebird-Architect@yahoogroups.com, "Dmitry Yemanov"
<dimitr@u...> wrote:
standard compliance is a three-line fix in dsql/gen.cpp I would
assume the same holds for FB2 and Vulcan.
Now we have to think about our installed base and how me might
migrate CHECK behaviour without leaving our beloved users up the
proverbial creek without tools.
Paul
--- In Firebird-Architect@yahoogroups.com, "Dmitry Yemanov"
<dimitr@u...> wrote:
><search
> "paulruizendaal" <pnr@w...> wrote:
> >
> > What are the opinions about this?
>
> The Firebird behaviour violates the standard.
>
> > What does the standard say?
>
> "A table check constraint is satisfied if and only if the specified
> condition> is not False for any row ofOracle's
> a table". And Unknown is not False.
>
> > What do other databases implement?
>
> PostgreSQL and MSSQL are standard compliant, i.e. they follow the
> line of thinking.
>
>
> Dmitry
>