Received-SPF: pass (sog-mx-3.v43.ch3.sourceforge.com: domain of bluematt.me
	designates 173.246.101.161 as permitted sender)
	client-ip=173.246.101.161;
	envelope-from=bitcoin-list@bluematt.me; helo=mail.bluematt.me; 
From: Matt Corallo <bitcoin-list@bluematt.me>
To: bitcoin-development@lists.sourceforge.net
In-Reply-To: <1323979147.27319.140661012141129@webmail.messagingengine.com>
References: <1323728469.78044.YahooMailNeo@web121012.mail.ne1.yahoo.com>
	<1323979147.27319.140661012141129@webmail.messagingengine.com>
Content-Type: text/plain; charset="UTF-8"
Date: Thu, 15 Dec 2011 21:48:17 -0500
Message-ID: <1324003697.3719.5.camel@BMThinkPad.lan.bluematt.me>
Mime-Version: 1.0
Content-Transfer-Encoding: 7bit
Subject: Re: [Bitcoin-development] [BIP 15] Aliases
Precedence: list

On Thu, 2011-12-15 at 13:59 -0600, theymos wrote:
> Bitcoin already has code and a protocol for transactions to IP
> addresses. Why not reuse that for dynamic address lookup? Just a few
> changes are necessary to enable complete user@server.com handling:
I'm not against this, but I think its way overcomplicated when compared
to the DNS or HTTPS methods.
> - Extend the protocol so that "reply" messages can be signed by a fixed
>   public key
> - Extend "checkorder" messages so they can specify an account to
>   send BTC to. Or standardize on how to put the account into the
>   message field.
OK, not too debatable, but considering how terrible bitcoind's account
handling is, the second might not be easy to get right...
> - Enable DNS lookups for IP transactions. The DNS-only proposals could
>   also be used here to avoid having to use the IP transaction protocol
>   sometimes. The public key for signing "reply" messages can be gotten
>   from TXT records. This will be safe with DNSSEC and Namecoin. With
>   plain DNS Bitcoin could take a SSH-like approach and ask the user to
>   verify the public key the first time it is used, remembering it later.
This is where I think this method becomes way overcomplicated.  Not only
do you have to update the IP-Transaction code, but now you have to
implement the full DNS System that is the other option as well.  Note
that to make this secure, we have to have a full DNSSEC-capable resolver
built-into bitcoind (there are libs, but it has to happen).  Yes you can
ask the user "does this fingerprint look right to you? Y/N" but that
always opens you up to a ton of users getting screwed out of coins and I
don't think it should be enabled, except in bitcoind, and since the main
target of this whole alias system is bitcoin-qt users, well...

Matt