Thread: RE: [SQL] random tuple

RE: [SQL] random tuple

From
"Jackson, DeJuan"
Date:
> On 07-Oct-98 Jackson, DeJuan wrote:
> >> Is there any way to select a random row from a table using an SQL
> >> query?
> >> I'm using Postgresql 6.3.2.
> >>
> > I'd look at using cursors and random().
>
> It's a good idea. Do you (or someone else) know how to use the
> following
> PostgreSQL functions: oidrand(oid,int4) ,oidsrand(int4)? What are they
> intended for and what is their result?
>
Never seen them before but it looks like:
    oidsrand(int4) -- seeds the random number generator for oidrand to
use
    oidrand(oid, int4) -- returns a psudo-random oid

The parameters to oidrand I can't figure out.
Anybody else?
        -DEJ
> ---
>
> ------------------------------------
>  Mauro Bartolomeoli
>  e-mail: mbarto@novacomp.it
>  ICQ#: 9602542
> ------------------------------------

Re: [HACKERS] RE: [SQL] random tuple

From
"Thomas G. Lockhart"
Date:
> > >> Is there any way to select a random row from a table?
> > > I'd look at using cursors and random().
> > Do you (or someone else) know how to use the following
> > PostgreSQL functions: oidrand(oid,int4) ,oidsrand(int4)?
>     oidsrand(int4) -- seeds the random number generator for oidrand
>     oidrand(oid, int4) -- returns a psudo-random oid

The regression test uses oidrand(), which is where I stumbled across it.
The behavior is that oidrand() returns a boolean true/false with an
inverse probability specified by the second argument. For example, given
a table t with 100 entries, and the query

  select * from t where oidrand(oid, 10)

will return, on average, 10 (10%) of the entries at random. The function
is called 100 times in the course of the query, and uses random() or
something similar to decide whether to return true or false for any
particular instance.

  select * from t where oidrand(oid, 1)

will, on average, return all entries (1/1 = 100%).

  select * from t where oidrand(oid, 100)

will, on average, return 1 entry (1/100 = 1%) so sometimes will return
one, zero, or two entries, and occasionally return more than two
entries.

It's pretty random, probably with a Poisson distribution depending on
what you are asking for.

Presumably oidsrand() allows one to change the seed to keep the
pseudo-random results from repeating from one run to the next. But I
haven't looked into it.

                   - Tom