trafodion-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Eric Owhadi <eric.owh...@esgyn.com>
Subject RE: [DISCUSS] Change the default setting of ALLOW_NULLABLE_UNIQUE_KEY_CONSTRAINT to ON
Date Tue, 02 Feb 2016 22:42:13 GMT
Can you elaborate on why you think the unique constraint can be broken
during execution? Not sure I understand why there is something special at
execution time. Once keys are set and are valid as unique, in what scenario
will they change and becomes non unique?
Eric

-----Original Message-----
From: Qifan Chen [mailto:qifan.chen@esgyn.com]
Sent: Tuesday, February 2, 2016 4:18 PM
To: dev <dev@trafodion.incubator.apache.org>
Subject: Re: [DISCUSS] Change the default setting of
ALLOW_NULLABLE_UNIQUE_KEY_CONSTRAINT to ON

The unique constraint probably could be broken during the execution.

For example, Probe-cache requires at most one row returned per probe and a
violation could mean more than one row is returned. Currently, the cache is
not built for such condition.

Thanks --Qifan

On Tue, Feb 2, 2016 at 4:09 PM, Eric Owhadi <eric.owhadi@esgyn.com> wrote:

> Can you elaborate on this?
> As I understand Suresh message, even with NULLABLE columns, we still
> assume uniqueness of the concatenation of all column making up the PK.
> So your sentence "a full PK will map to up to one row" is still respected?
> Or am I misunderstanding?
> Eric
>
> -----Original Message-----
> From: Qifan Chen [mailto:qifan.chen@esgyn.com]
> Sent: Tuesday, February 2, 2016 4:05 PM
> To: dev <dev@trafodion.incubator.apache.org>
> Subject: Re: [DISCUSS] Change the default setting of
> ALLOW_NULLABLE_UNIQUE_KEY_CONSTRAINT to ON
>
> Allowing NULL in primary key columns could break some existing
> assumptions in our code base on PKs. For example, a full PK will map
> to up to one row, which is a logic constraint that the compiler takes full
> advantage of.
>
> Thanks --Qifan
>
> On Tue, Feb 2, 2016 at 4:01 PM, Dave Birdsall
> <dave.birdsall@esgyn.com>
> wrote:
>
> > ANSI in general does not concern itself with physical issues. So no
> > partitioning, no indexing, and so on in ANSI.
> >
> >
> > -----Original Message-----
> > From: Eric Owhadi [mailto:eric.owhadi@esgyn.com]
> > Sent: Tuesday, February 2, 2016 1:59 PM
> > To: dev@trafodion.incubator.apache.org
> > Subject: RE: [DISCUSS] Change the default setting of
> > ALLOW_NULLABLE_UNIQUE_KEY_CONSTRAINT to ON
> >
> > Great, so my 2 cts is that we should not worry about ANSI on the
> > default not supporting NULL, since ANSI did not account for
> > partitioning...
> > Eric
> >
> > -----Original Message-----
> > From: Roberta Marton [mailto:roberta.marton@esgyn.com]
> > Sent: Tuesday, February 2, 2016 3:55 PM
> > To: dev@trafodion.incubator.apache.org
> > Subject: RE: [DISCUSS] Change the default setting of
> > ALLOW_NULLABLE_UNIQUE_KEY_CONSTRAINT to ON
> >
> > Store by is not ANSI standard.
> >
> >    Roberta
> >
> > -----Original Message-----
> > From: Eric Owhadi [mailto:eric.owhadi@esgyn.com]
> > Sent: Tuesday, February 2, 2016 1:53 PM
> > To: dev@trafodion.incubator.apache.org
> > Subject: RE: [DISCUSS] Change the default setting of
> > ALLOW_NULLABLE_UNIQUE_KEY_CONSTRAINT to ON
> >
> > Do you happen to know if STORE BY is part of the ANSI standard, and
> > if ANSI mandate that anything in STORE BY must be part of the PK?
> > Eric
> >
> > -----Original Message-----
> > From: Dave Birdsall [mailto:dave.birdsall@esgyn.com]
> > Sent: Tuesday, February 2, 2016 3:45 PM
> > To: dev@trafodion.incubator.apache.org
> > Subject: RE: [DISCUSS] Change the default setting of
> > ALLOW_NULLABLE_UNIQUE_KEY_CONSTRAINT to ON
> >
> > Our default behavior would not be ANSI standard if this is changed.
> > I don’t know if that is a compelling reason not to do this though.
> >
> > Dave
> >
> > -----Original Message-----
> > From: Suresh Subbiah [mailto:suresh.subbiah60@gmail.com]
> > Sent: Tuesday, February 2, 2016 1:37 PM
> > To: dev@trafodion.incubator.apache.org
> > Subject: [DISCUSS] Change the default setting of
> > ALLOW_NULLABLE_UNIQUE_KEY_CONSTRAINT to ON
> >
> > Hi,
> >
> > Do you foresee potential problems if  we change the default setting
> > for ALLOW_NULLABLE_UNIQUE_KEY_CONSTRAINT to ON. ?
> > It does seem that users are asking for using and setting this CQD in
> > their individual applications.
> > Other than JIRA 1801, there are no known issues when this attribute
> > is set to ON.
> > Having the default set to ON could make Trafodion easier to use.
> >
> > I can file a JIRA if there is agreement or if we think the
> > discussion should be moved to a JIRA.
> >
> > Thank you
> > Suresh
> >
>
>
>
> --
> Regards, --Qifan
>



--
Regards, --Qifan

Mime
View raw message