Re: OID wraparound (was Re: pg_depend) - Mailing list pgsql-hackers

From Horst Herb
Subject Re: OID wraparound (was Re: pg_depend)
Date
Msg-id 01071913032800.02053@munin.gnumed.dhs.org
Whole thread Raw
In response to Re: OID wraparound (was Re: pg_depend)  (Bruce Momjian <pgman@candle.pha.pa.us>)
List pgsql-hackers
On Thursday 19 July 2001 06:08, you wrote:
> > Bruce Momjian <pgman@candle.pha.pa.us> writes:

> I think it should be off on user tables by default, but kept on system
> tables just for completeness.  It could be added at table creation time
> or from ALTER TABLEL ADD.  It seems we just use them too much for system
> stuff.  pg_description is just one example.

and what difference should it make, to have a few extra hundred or thousand 
OIDs used by system tables, when I insert daily some ten thousand records 
each using an OID for itself?

Why not make OIDs 64 bit? Might slow down a little on legacy hardware, but in 
a couple of years we'll all run 64 bit hardware anyway.

I believe that just using 64 bit would require the least changes to Postgres. 
Now, why would that look that obvious to me and yet I saw no mentioing of 
this in the recent postings. Surely it has been discussed before, so which is 
the point I miss or don't understand?

I would need 64 bit sequences anyway, as it is predictable that our table for 
pathology results will run out of unique IDs in a couple of years.

Horst 


pgsql-hackers by date:

Previous
From: Justin Koivisto
Date:
Subject: Re: All computers in the world MUST sync with ATOMIC clock before 12:00 AM 21 July 2001!!!
Next
From: "August Zajonc"
Date:
Subject: Neutral Soil (OT)