sequences

Started by Brett McCormickalmost 28 years ago6 messages
#1Brett McCormick
brett@work.chicken.org

Are there any plans to implement oid-style functionality for
sequences, such as returning the nextval in the result tuple somehow?

Is it at all possible to implement? That is one very nice thing about
using oid instead, is that it comes back in PQoidStatus. I probably
won't stop using oids until this is possible (i've got my own restore
which "preserves" (i.e. renumbers) oid references)

--brett

#2Vadim B. Mikheev
vadim@sable.krasnoyarsk.su
In reply to: Brett McCormick (#1)
Re: [HACKERS] sequences

Brett McCormick wrote:

Are there any plans to implement oid-style functionality for
sequences, such as returning the nextval in the result tuple somehow?

Is it at all possible to implement? That is one very nice thing about
using oid instead, is that it comes back in PQoidStatus. I probably
won't stop using oids until this is possible (i've got my own restore
which "preserves" (i.e. renumbers) oid references)

Not possible - sequences are just "a gear", user can use more than
one sequences in inserts, etc...

But we could return PRIMARY KEY of affected tuple if it exists!
I kept this feature in mind when thought about removing OID...

Comments ?

Vadim

#3Michael J. Rogan
mrogan@fpelectronics.com
In reply to: Vadim B. Mikheev (#2)
Re: [HACKERS] sequences

Brett McCormick wrote:

Are there any plans to implement oid-style functionality for
sequences, such as returning the nextval in the result tuple somehow?

Is it at all possible to implement? That is one very nice thing about
using oid instead, is that it comes back in PQoidStatus. I probably
won't stop using oids until this is possible (i've got my own restore
which "preserves" (i.e. renumbers) oid references)

Not possible - sequences are just "a gear", user can use more than
one sequences in inserts, etc...

But we could return PRIMARY KEY of affected tuple if it exists!
I kept this feature in mind when thought about removing OID...

Comments ?

This would be very useful. I never did like the idea of using OID as
reference so a way to get the PIMARY KEY would be A GOOD THING(tm).

Michael

BTW: I could have sworn the first thing I tried was to grant select on the
sequence I was using when it complianed about access.

Vadim

* Michael J. Rogan, Network Administrator, 905-624-3020 *
* Mark IV Industries, F-P Electronics & I.V.H.S. Divisions *
* mrogan@fpelectronics.com mrogan@ivhs.com *

#4Vadim B. Mikheev
vadim@sable.krasnoyarsk.su
In reply to: Michael J. Rogan (#3)
Re: [HACKERS] sequences

Michael J. Rogan wrote:

But we could return PRIMARY KEY of affected tuple if it exists!
I kept this feature in mind when thought about removing OID...

Comments ?

This would be very useful. I never did like the idea of using OID as
reference so a way to get the PIMARY KEY would be A GOOD THING(tm).

But we will have to decide how to deal with multi-key PK...

Vadim

#5Brett McCormick
brett@work.chicken.org
In reply to: Vadim B. Mikheev (#4)
Re: [HACKERS] sequences

On Fri, 27 February 1998, at 10:27:07, Vadim B. Mikheev wrote:

Michael J. Rogan wrote:

But we could return PRIMARY KEY of affected tuple if it exists!
I kept this feature in mind when thought about removing OID...

Comments ?

This would be very useful. I never did like the idea of using OID as
reference so a way to get the PIMARY KEY would be A GOOD THING(tm).

But we will have to decide how to deal with multi-key PK...

Vadim

How is this implemented?

#6Vadim B. Mikheev
vadim@sable.krasnoyarsk.su
In reply to: Michael J. Rogan (#3)
Re: [HACKERS] sequences

Brett McCormick wrote:

On Fri, 27 February 1998, at 10:27:07, Vadim B. Mikheev wrote:

Michael J. Rogan wrote:

But we could return PRIMARY KEY of affected tuple if it exists!
I kept this feature in mind when thought about removing OID...

Comments ?

This would be very useful. I never did like the idea of using OID as
reference so a way to get the PIMARY KEY would be A GOOD THING(tm).

But we will have to decide how to deal with multi-key PK...

Vadim

How is this implemented?

Using multi-key unique btree indices.

Vadim