Date: Thu, 15 Aug 2013 10:37:42 +0200
From: Pieter Wuille <pieter.wuille@gmail.com>
To: Mike Hearn <mike@plan99.net>
Message-ID: <20130815083741.GA12713@vps7135.xlshosting.net>
References: <CABsx9T3DM72+8HgNWWZ2HaAgZMQGAPn87L9VVKdkbVkS7sd8Tg@mail.gmail.com>
	<CANEZrP3bt40ThBcqWLfzuS5508mpbo4Z5qxh9AJs-FRrk0QJsA@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <CANEZrP3bt40ThBcqWLfzuS5508mpbo4Z5qxh9AJs-FRrk0QJsA@mail.gmail.com>
User-Agent: Mutt/1.5.21 (2010-09-15)
Cc: Bitcoin Dev <bitcoin-development@lists.sourceforge.net>
Subject: Re: [Bitcoin-development] Version 0.9 goals
Precedence: list

On Thu, Aug 15, 2013 at 10:09:48AM +0200, Mike Hearn wrote:
> Sounds awesome!
> 
> Pieter told me at lunch that headers first cut sync time to 45 minutes for
> him, which is another amazing improvement from the master of optimisations.

Just to make sure nobody expects a magic bullet: this was on a hexacore Xeon
CPU, with several GB of -dbcache, libsecp256k1 for verification, and a very good
network connection. It is repeatable and from random network peers, though. The
code is here:

  https://github.com/sipa/bitcoin/commits/headersfirst

It's usable and seems to be stable (including reindexing, which needs support for
block files with out-of-order blocks now), but I still want to clean some
things up before pullreq'in. There are probably some heuristic tweaks
possible as well - Gregory found that performance is reduced for the first
part of the chain on high-latency networks.

> Pieter, Matt and I also agreed that for maximum impact we should really try
> to ship payment protocol support in at least two clients simultaneously and
> ideally with a big merchant signed up too - to send a powerful message that
> we really mean it. Someone volunteered last week to do it for bitcoinj and
> if he doesn't pull through, I have some old code from EOY 2012 that I could
> update to the latest spec and ship at least some basic support. I'd hope
> that we can get Bitcoin Wallet or MultiBit updates out once bcj has support
> pretty fast.
> 
> Also, Jeff said that BitPay want to be a leader in support for the
> protocol. So let's try and co-ordinate release dates so we can make a bit
> of a splash and grab the ecosystems attention.

I believe we do need some wider support than just Bitcoin-Qt, indeed, as
the number of people actually using the reference client as a wallet is
quite low now. Ideally, several clients and merchants start support for it
in a short timeframe...

-- 
Pieter