| From: | tomas(at)tuxteam(dot)de |
|---|---|
| To: | Hannu Krosing <hannu(at)skype(dot)net> |
| Cc: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>, Alvaro Herrera <alvherre(at)commandprompt(dot)com>, "Jim C(dot) Nasby" <jim(at)nasby(dot)net>, Warren Turkal <wt(at)penguintechs(dot)org>, pgsql-hackers(at)postgresql(dot)org |
| Subject: | Re: RFC: Temporal Extensions for PostgreSQL |
| Date: | 2007-02-19 05:45:47 |
| Message-ID: | 20070219054547.GA27947@www.trapp.net |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
> Well, unique is usually defined as "not equal to any other". And "not
> equal" also fails transitive law [...]
> But it should be trivial to test at insertion time if the interval
> overlaps with any existing intervals [...]
Putting your point another way: you might construe an equivalence
relation by grouping together all intervals which (directly or
indirectly) touch each other. Let's say they are "connected".
But then the problem becomes clear: let's assume A and C are not
connected (i.e. they are in different equivalence classes). Now you add
B, which happens to overlap A and C. Now A and C are connected. How do
you care for that in your index?
That can't happen with a "classical" equivalence relation, which
wouldn't change among existing elements when you add a new one.
Regards
- -- tomás
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (GNU/Linux)
iD8DBQFF2TmLBcgs9XrR2kYRAmIHAJ4+x1mOum1rvBkS8/Pypcu8w2QIIQCffFm5
No5aOh901rxfc2mpRYpJMAU=
=7Isi
-----END PGP SIGNATURE-----
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Greg Smith | 2007-02-19 07:38:09 | Re: WIP patch - INSERT-able log statements |
| Previous Message | Oleg Bartunov | 2007-02-19 04:47:03 | Re: n-gram search function |