Date: Mon, 6 Aug 2018 18:39:25 +1000
From: Anthony Towns <aj@erisian.com.au>
To: Russell O'Connor <roconnor@blockstream.io>,
	Bitcoin Protocol Discussion <bitcoin-dev@lists.linuxfoundation.org>
Message-ID: <20180806083925.kg5px476bzhec44b@erisian.com.au>
References: <CAPg+sBj7f+=OYXuOMdNeJk3NBG67FSQSF8Xv3seFCvwxCWq69A@mail.gmail.com>
	<A899D97B-5D47-4AB0-8A7F-57F91C58ADE1@sprovoost.nl>
	<CAPg+sBg1WuG1MihC3zBHJpxVqC2Sys7Y52iWs6JXEMmnL_tE_w@mail.gmail.com>
	<CAMZUoKm4Qs2yAc+WKgN1J2D8MDgbzNnK69kF+hbY2GDyRqdVdg@mail.gmail.com>
	<CAMZUoKm_ij4Ffzx5Wpipa5RAFA=5F06jhiTCMJhp3vAj1q+2jA@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <CAMZUoKm_ij4Ffzx5Wpipa5RAFA=5F06jhiTCMJhp3vAj1q+2jA@mail.gmail.com>
User-Agent: NeoMutt/20170113 (1.7.2)
Subject: Re: [bitcoin-dev] Schnorr signatures BIP
Precedence: list

On Sun, Aug 05, 2018 at 10:33:52AM -0400, Russell O'Connor via bitcoin-dev wrote:
> In light of this, I revise my proposed change to make the verification
> equation
> 
> R + sG + eP = 0.

Isn't the verification equation "R + s(-G) + eP = 0" equally good, then,
since -G is a constant? (ie, at worst it's a matter of optimising the
verifier for -G as well as G)

If not, what's the actual performance impact of having to negate "s"
as part of batch verifying ~10000 signatures? It seems like it should
be trivially small to me? (scalar_negate benchmarks at 0.00359us, while
ecdsa_verify benchmarks at 66us, which I believe then reduces by a factor
of ~3 for batches of 10k schnorr sigs?)

FWIW, I'm a fan of the formulation "s = r + H(R,P,m)p" mostly because
it seems like the simplest possible way of describing the setup, and I'm
all for optimising for people being able to understand what's going on.

Cheers,
aj