Received: from localhost.localdomain (postgresql.org [64.49.215.8]) by localhost (Postfix) with ESMTP id 53B324761F3; Tue, 21 May 2002 08:53:19 -0400 (EDT) Received: from pepe.lfcity.org (sanpedro-a438.racsa.co.cr [196.40.41.186]) by postgresql.org (Postfix) with ESMTP id 7C45B475DED; Tue, 21 May 2002 08:52:47 -0400 (EDT) Received: from there (localhost.localdomain [127.0.0.1]) by pepe.lfcity.org (Postfix) with SMTP id C87C780269; Tue, 21 May 2002 06:56:16 -0500 (CDT) Content-Type: text/plain; charset="iso-8859-1" From: andre To: Bruno Wolff III Subject: Re: Guarantee uniqueness of PGP key Date: Tue, 21 May 2002 06:56:16 -0500 X-Mailer: KMail [version 1.3.1] Cc: pgsql-sql@postgresql.org, pgsql-novice@postgresql.org References: <20020520201709.A3F8A80269@pepe.lfcity.org> <20020521115442.GA17061@wolff.to> In-Reply-To: <20020521115442.GA17061@wolff.to> MIME-Version: 1.0 Content-Transfer-Encoding: quoted-printable Message-Id: <20020521115616.C87C780269@pepe.lfcity.org> X-Archive-Number: 200205/218 X-Sequence-Number: 3848 Yes, I've decided to relax the unique-ness constraint on the public key. T= he=20 email address is still guaranteed unique. And we send out an encrypted=20 verification email that only the person with the private key can decrypt an= d=20 respond to. Thus, it doesn't really hurt to have multiple people paste in t= he=20 same pub key. If I really wanted uniqueness, I think you (and the others who replied) are= =20 correct that I should be storing the 'fingerprint' in a separate column. /andre=20=20 On Tuesday 21 May 2002 06:54 am, Bruno Wolff III wrote: > On Mon, May 20, 2002 at 15:17:09 -0500, > > andre wrote: > > The problem is that some PGP keys are over 2713 bytes, and postgres (7.= 2) > > gives the error: > > You might want to rethink your design. You probably want just the RSA or > DSA public key to be unique. Otherwise you will treat different versions = of > what is essentially the same key as being different. (This might be due > to be signed by a different set of keys.) I suspect that this isn't what > you want. I think even the craziest people are using keys that should > be around 1000 bytes. Paranoid people (or those needing very long lived > public keys) are using ones half that size.