From: Rusty Russell <rusty@rustcorp.com.au>
To: <bitcoin-dev@lists.linuxfoundation.org>
Date: Fri, 10 Nov 2017 12:01:14 +1030
Message-ID: <87lgjex6l9.fsf@rustcorp.com.au>
MIME-Version: 1.0
Content-Type: text/plain
Cc: Christian Decker <decker@blockstream.com>,
	Russell O'Connor <roconnor@blockstream.com>
Subject: [bitcoin-dev] Covenants through merkelized txids.
Precedence: list

Hi all,

        This is an alternative to jl2012's BIPZZZ (OP_PUSHTXDATA[1]).
It riffs on the (ab)use of OP_CHECKSIGFROMSTACK that Russell[2]
used to implement covenants[3].  I've been talking about it to random
people for a while, but haven't taken time to write it up.

The idea is to provide a OP_TXMERKLEVERIFY that compares the top stack
element against a merkle tree of the current tx constructed like so[4]:

        TXMERKLE = merkle(nVersion | nLockTime | fee, inputs & outputs)
        inputs & outputs = merkle(inputmerkle, outputmerkle)
        input = merkle(prevoutpoint, nSequence | inputAmount)
        output = merkle(nValue, scriptPubkey)

Many variants are possible, but if we have OP_CAT, this makes it fairly
easy to test a particular tx property.  A dedicated OP_MERKLE would make
it even easier, of course.

If we one day HF and add merklized TXIDs[5], we could also use this method
to inspect the tx *being spent* (which was what I was originally trying to
do).

Thanks for reading!
Rusty.

[1] https://github.com/jl2012/bips/blob/vault/bip-0ZZZ.mediawiki
[2] Aka Dr. "Not Rusty" O'Connor.  Of course both of us in the same thread will
    probably break the internet.
[3] https://blockstream.com/2016/11/02/covenants-in-elements-alpha.html
[4] You could put every element in a leaf, but that's less compact to
    use: cheaper to supply the missing parts with OP_CAT than add another level.
[5] Eg. use the high nVersion bit to say "make my txid a merkle".