Received-SPF: pass (sog-mx-4.v43.ch3.sourceforge.com: domain of taplink.co
	designates 50.117.27.232 as permitted sender)
	client-ip=50.117.27.232; envelope-from=jeremy@taplink.co;
	helo=mail.taplink.co; 
Content-Type: text/plain; charset=iso-8859-15; format=flowed; delsp=yes
To: "Gregory Maxwell" <gmaxwell@gmail.com>, "Mike Hearn" <mike@plan99.net>,
	"Matt Corallo" <bitcoin-list@bluematt.me>
References: <52A3C8A5.7010606@gmail.com>
	<1795f3067ba3fcdd0caf978cc59ff024.squirrel@fruiteater.riseup.net>
	<52A435EA.7090405@gmail.com> <201312081237.24473.luke@dashjr.org>
	<CANAnSg2OrmQAcZ+cZdtQeADicH3U29QOgYPfP1AQhOMP6+P1wg@mail.gmail.com>
	<CAAS2fgR0khyJxmz9c2Oc87hOFgiNuiPJuaeugGajdo_EcKEW9w@mail.gmail.com>
	<20131212205106.GA4572@netbook.cypherspace.org>
	<CANAnSg3nPhrk2k=yDKf39AuBQnSuTWJbgANdMhGe=soiOy0NTw@mail.gmail.com>
	<CAAS2fgTmWRMxYweu3sNn_X7grgjUqTQujM-DbZRxG_YMZnD=7g@mail.gmail.com>
	<CANEZrP2X_63qkuNuk0MvsLR9ewd7HR0mPVaD7bZSgWMTJ5-9FQ@mail.gmail.com>
	<CAAS2fgQmMZ6RYjbJ6ZfFO5+ZhZoR4d4yMf8CXLXKPmZt3-Je4Q@mail.gmail.com>
	<CANEZrP1mdJNa7ADkUXTGDNKMSCROjGAVbMXZXxodxCz1LFDzZw@mail.gmail.com>
	<op.w8y642nryldrnw@laptop-air.hsd1.ca.comcast.net>
	<4264e886-48de-40ac-921a-a60502595060@email.android.com>
Date: Wed, 01 Jan 2014 02:02:02 -0800
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
From: "Jeremy Spilman" <jeremy@taplink.co>
Organization: TapLink
Message-ID: <op.w8z55onhyldrnw@laptop-air.hsd1.ca.comcast.net>
In-Reply-To: <4264e886-48de-40ac-921a-a60502595060@email.android.com>
User-Agent: Opera Mail/1.0 (Win32)
oclient: 192.168.168.135#jeremy@taplink.co#465
Cc: Bitcoin Dev <bitcoin-development@lists.sourceforge.net>
Subject: Re: [Bitcoin-development] Dedicated server for bitcoin.org,
	your thoughts?
Precedence: list

So I looked into gitian, the first thing I noticed was the hashes that  
people were signing, for example:

    https://github.com/bitcoin/gitian.sigs/blob/master/0.8.6-win32/gavinandresen/bitcoin-build.assert

don't match the hash of the file 'bitcoin-0.8.6-win32-setup.exe' actually  
hosted by sourceforce. That was a bit alarming at first, but I talked to  
BlueMatt and maaku on IRC and the difference is due to Gavin Authenticode  
signing the executable for windows.

BlueMatt asked if someone could implement in gitian-downloader a way to  
strip off the signature so that we could get back to the raw binary with a  
hash that matches what everyone is producing from gitian.  I found:

   http://blog.didierstevens.com/programs/disitool/

which is a Python script which can strip the signature nicely, but the  
hashes still don't match.

I couldn't find a gitian build of 0.8.6 so I built my own, and after  
verifying the hash for v0.8.6 was '49547ff9...' as expected I looked at  
the hex diff between that and the sig-stripped .exe from sourceforge, and  
the only two differences are:

   - At offset D8 the stripped file has '5D E2 B2' versus 'F9 F4 00' in the  
gitian build
   - The sig-stripped file has an extra byte '00' at the end

I started to look at the file spec for windows PE files and quickly  
thought better of it. Maybe someone better informed can chime in on what  
those three bytes at offset D8 specify.

I'm not sure if we want to patch the signature onto the gitian build, or  
strip the signature off of the Gavin-signed build, but something of the  
sort is necessary if you want get gitian-downloader to match the official  
distro (for Windows at least).

In any case, I think wallet users want to know when an upgrade is  
available, and ability to click an 'update' button get a binary they can  
trust. It's not a problem unique to bitcoind, deterministic builds are  
awesome, but I don't think fully solve it.

Thanks,
Jeremy

On Tue, 31 Dec 2013 13:33:54 -0800, Matt Corallo  
<bitcoin-list@bluematt.me> wrote:

> We already have a wonderful system for secure updating -  
> gitian-downloader. We just neither use it >not bother making actual  
> gitian releases so anyone can use it to verify signatures of downloads.