| From: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
|---|---|
| To: | elein(at)norcov(dot)com |
| Cc: | pgsql-general(at)postgresql(dot)org |
| Subject: | Re: Fwd: Sizes of sequences and serials |
| Date: | 2002-09-27 03:56:23 |
| Message-ID: | 20713.1033098983@sss.pgh.pa.us |
| Views: | Whole Thread | Raw Message | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-general |
elein <elein(at)sbcglobal(dot)net> writes:
> According to the documentation, it looks like sequences
> are always stored as bigints (if the platform supports them).
Correct.
> Serials, though, seem to be able to be declared as serial
> with integer storage or serial8 for bigint storage.
> Is the storage for ordinary serials really an int and not
> a big int?
If you declare a serial (aka serial4) column, you get an int4 column
that's driven by a bigint sequence. No problem except for a runtime
conversion from int8 to int4, which is really pretty negligible in
comparison to the cost of nextval().
> Is there a way to create a sequence stored as an int instead of a big
> int?
No; it wouldn't save anything anyway, seeing that each sequence is
stored as a separate table and so must occupy one disk block.
regards, tom lane
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Bruce Momjian | 2002-09-27 03:58:20 | Re: postmaster -d option (was Re: [GENERAL] Relation 0 does |
| Previous Message | Alvaro Herrera | 2002-09-27 03:31:06 | Re: get date in binary number format |