From: | "Patrick Earl" <patearl(at)patearl(dot)net> |
---|---|
To: | pgsql-general(at)postgresql(dot)org |
Subject: | GUID/UUID Support |
Date: | 2007-01-16 22:47:09 |
Message-ID: | e0762e610701161447q537f7492iee797f82fdb14a64@mail.gmail.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-general |
Greetings all. I've been poking around the web and trying to track
the status of the GUID/UUID patches. In the todo list, that item is
not marked as going into 8.3, and I feel this is unfortunate.
Searching google for "postgresql guid -guide" comes up with around
200,000 hits. It's obviously been a popular topic over time.
There was some discussion about including a generator in the core or
not. Even a random generator (version 4 UUID) would be great in my
opinion. The Wikipedia article about UUIDs produces some statistics
about a random source ID that are interesting.
http://en.wikipedia.org/wiki/UUID
If you generate 2^41 (2'199'023'255'552) true random ids, the
probability of getting a duplicate is 0.000'000'000'000'4. For
comparison, the page indicates that the probability of getting hit by
a meteorite is 0.000'000'000'06.
If there are qualms about a generator that only produces one type of
IDs, could the database not just provide a way to generate a "version
X" uuid? Some versions would be implemented later.
One thing that is quite important, to me at least, is to make the GUID
generation functionality cross-platform. Using a random UUID
technique for an initial implementation would simplify this issue.
Even if the generator is punted to contrib, I would like to see at
least native UUID field support in the next version.
Patrick
From | Date | Subject | |
---|---|---|---|
Next Message | Chad Wagner | 2007-01-16 23:30:00 | Re: invalid byte sequence for encoding "UTF8" |
Previous Message | Martijn van Oosterhout | 2007-01-16 22:04:55 | Re: 8.1 vs 8.2.1 view optimization |