Date: Thu, 29 Dec 2011 12:01:20 -0500 (EST)
From: roconnor@theorem.ca
To: Gavin Andresen <gavinandresen@gmail.com>
In-Reply-To: <CABsx9T06H29R4CpL9hXF_yyB4chko1YdkhbCZ8rdwo1gLmF1BQ@mail.gmail.com>
Message-ID: <alpine.LRH.2.00.1112291143220.22327@theorem.ca>
References: <alpine.LRH.2.00.1112290111310.22327@theorem.ca>
	<CABsx9T06H29R4CpL9hXF_yyB4chko1YdkhbCZ8rdwo1gLmF1BQ@mail.gmail.com>
User-Agent: Alpine 2.00 (LRH 1167 2008-08-23)
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII; format=flowed
Cc: bitcoin-development@lists.sourceforge.net, pool@deepbit.net,
	webmaster@btcguild.com
Subject: Re: [Bitcoin-development] Alternative to OP_EVAL
Precedence: list

Good morning everyone.

On Thu, 29 Dec 2011, Gavin Andresen wrote:

> First, thanks very much to Russell for looking more closely at both
> BIP 12 and the patch than anybody else-- he's found two bugs and two
> things the BIP isn't clear enough on (so far).
>
> And I've got to say, I'm very sympathetic to the "OP_EVAL starts down
> the code-as-data path, and There Be Dragons" argument.
>
> But:
>
> I don't think the proposed alternative would be, in practice, any
> better.  I see two main disadvantages over OP_EVAL:
>
>  about 20-bytes larger
>
>  it means going back to where we were two months ago, writing more
> code, reviewing it, finding bugs in it, backporting it so miners
> running old software can support it, etc.

Well, given the state that the OP_EVAL proposal was in when I looked at it 
this week, all your code reviews you have done so far are not adequate 
anyways.

Gavin, push the OP_EVAL date back 2 months.  OP_EVAL just is not ready 
yet.

> ... and some other minor disadvantages:
>
>  'standard' scripts will need to be slightly different in the
> scriptSig and the scriptPubKey
>   (e.g. <signature> CHECKSIG  becomes  <signature> CHECKSIGVERIFY
> with OP_CODEHASH)
>
>  OP_EVALs are not executed, and so the code associated with them does
> not have to be part of the transaction, if they are in the
> non-executed branch of an OP_IF. That could be good for privacy, and
> could be good for reducing block-chain size.

I don't understand the above paragraph.

> ----------------------
>
> In discussions in IRC yesterday, we talked a little about possible
> changes to the OP_EVAL BIP to make it less subject to abuse. In
> particular, the big can of worms is allowing arithmetic or bit
> operations on the serialized script that will be EVAL'ed:
>  <serialized script> <other_data> OP_ADD OP_EVAL  <-- Look! Dragons!
>
> If <serialized script> is more than 4 bytes, that is actually illegal
> right now (all of the arithmetic operations are limited to operating
> on numbers that are 4 bytes of less, and I believe we could prove that
> no series of operations will ever produce a value more than 5 bytes
> big given the current limitations).

This is not adequate: <data> OP_SHA256 OP_EVAL runs random code that is 
more than 5 bytes.

> Which leads me to suggest that BIP 12 be amended to state that:
>  OP_EVAL shall cause script validation to fail if the top item on the
> stack is less than 8 bytes long.
>
> I'm tempted to propose a rule:
>  OP_EVAL shall fail if the top item on the stack is the result of any
> calculation
>
> ... but I don't think the extra code it would take to implement that
> (keep track of which items on the stack were the results of
> OP_ADD/etc) is worth it.
>
> On the "you can't tell how many CHECKSIG operations will be performed
> before executing the script" issue:
>
> That is already true, because the parameters to CHECKMULTISIG that
> determine how many signatures it checks might be computed.

Yes, but maybe there is other static analysis miners may want to do.  I 
can't imagine every scenario.

> Finally, I would echo theymos' observation that I think we'll
> eventually do something very much like OP_EVAL in the future-- maybe
> to support (in a backwards-compatible way) a
> quantum-computing-resistant signature algorithm or SHA3. When that is
> done, I think it might make sense to do a bottom-up redesign of Script
> based on what we've learned.

IMHO I think the above observation is not very relevant to the merits of 
the existing OP_EVAL proposal on the table.

-- 
Russell O'Connor                                      <http://r6.ca/>
``All talk about `theft,''' the general counsel of the American Graphophone
Company wrote, ``is the merest claptrap, for there exists no property in
ideas musical, literary or artistic, except as defined by statute.''