From: | "David Favro" <postgres(at)meta-dynamic(dot)com> |
---|---|
To: | pgsql-general(at)lists(dot)postgresql(dot)org |
Subject: | 'Identifier' columns |
Date: | 2018-08-14 02:19:30 |
Message-ID: | 20180814022005.E2B66FB1@mail.meta-dynamic.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-general |
A couple of questions about auto-assigned identifier columns,
forgive my ignorance, I'm used to other methods to create IDs...
1. If creating a new application [i.e. no "legacy" reasons to do
anything] using PostgreSQL 10, when creating an "auto-assigned
integer ID" column, what are the advantages/disadvantages of using
the 'SERIAL' datatype [or equivalent explicitly created SEQUENCE w/
nextval() used as default for column] versus the SQL-standard
'integer GENERATED AS IDENTITY'? All other things being equal, it
would seem a no-brainer to follow the standard.
2. When using the SQL-standard 'integer GENERATED AS IDENTITY'
column, after inserting a column, what is the recommended method to
find the ID of the just-inserted row? Is there no SQL-standard way?
The docs seem to imply (without explicitly stating) that a SEQUENCE
is used behind the scenes hence 'currval()' could be used, but I
didn't see in the docs any mention of what the underlying sequence's
name is, or how to specify a name. Perhaps 'lastval()' would work,
but not in all cases and in any event it has a sloppy feel to me.
Thank you in advance for any advice that can be offered.
-- David
From | Date | Subject | |
---|---|---|---|
Next Message | Laurenz Albe | 2018-08-14 07:17:27 | Re: 'Identifier' columns |
Previous Message | Vick Khera | 2018-08-13 22:55:09 | Re: Vacuum process waiting on BufferPin |