[bitcoin-dev] Segregated Witness features wish list

[jl2012@xbt.hk]

It seems the current consensus is to implement Segregated Witness. SW 
opens many new possibilities but we need a balance between new features 
and deployment time frame. I'm listing by my priority:

1-2 are about scalability and have highest priority

1. Witness size limit: with SW we should allow a bigger overall block 
size. It seems 2MB is considered to be safe for many people. However, 
the exact size and growth of block size should be determined based on 
testing and reasonable projection.

2. Deployment time frame: I prefer as soon as possible, even if none of 
the following new features are implemented. This is not only a technical 
issue but also a response to the community which has been waiting for a 
scaling solution for years

3-6 promote safety and reduce level of trust (higher priority)

3. SIGHASH_WITHINPUTVALUE [1]: there are many SIGHASH proposals but this 
one has the highest priority as it makes offline signing much easier.

4. Sum of fee, sigopcount, size etc as part of the witness hash tree: 
for compact proof of violations in these parameters. I prefer to have 
this feature in SWv1. Otherwise, that would become an ugly softfork in 
SWv2 as we need to maintain one more hash tree

5. Height and position of an input as part of witness will allow compact 
proof of non-existing UTXO. We need this eventually. If it is not done 
in SWv1, we could softfork it nicely in SWv2. I prefer this earlier as 
this is the last puzzle for compact fraud proof.

6. BIP62 and OP_IF malleability fix [2] as standardness rules: 
involuntary malleability may still be a problem in the relay network and 
may make the relay less efficient (need more research)

7-15 are new features and long-term goals (lower priority)

7. Enable OP_CAT etc:
OP_CAT will allow tree signatures described by [3]. Even without Schnorr 
signature, m-of-n multisig will become more efficient if m < n.

OP_SUBSTR/OP_LEFT/OP_RIGHT will allow people to shorten a payment 
address, while sacrificing security.

I'm not sure how those disabled bitwise logic codes could be useful

Multiplication and division may still considered to be risky and not 
very useful?

8. Schnorr signature: for very efficient multisig [3] but could be 
introduced later.

9. Per-input lock-time and relative lock-time: define lock-time and 
relative lock-time in witness, and signed by user. BIP68 is not a very 
ideal solution due to limited lock time length and resolution

10. OP_PUSHLOCKTIME and OP_PUSHRELATIVELOCKTIME: push the lock-time and 
relative lock-time to stack. Will allow more flexibility than OP_CLTV 
and OP_CSV

11. OP_RETURNTURE which allows softfork of any new OP codes [4]. It is 
not really necessary with the version byte design but with OP_RETURNTURE 
we don't need to pump the version byte too frequently.

12. OP_EVAL (BIP12), which enables Merkleized Abstract Syntax Trees 
(MAST) with OP_CAT [5]. This will also obsolete BIP16. Further 
restrictions should be made to make it safe [6]:
a) We may allow at most one OP_EVAL in the scriptPubKey
b) Not allow any OP_EVAL in the serialized script, nor anywhere else in 
the witness (therefore not Turing-complete)
c) In order to maintain the ability to statically analyze scripts, the 
serialized script must be the last push of the witness (or script 
fails), and OP_EVAL must be the last OP code in the scriptPubKey

13. Combo OP codes for more compact scripts, for example:

OP_MERKLEHASH160, if executed, is equivalent to OP_SWAP OP_IF OP_SWAP 
OP_ENDIF OP_CAT OP_HASH160 [3]. Allowing more compact tree-signature and 
MAST scripts.

OP_DUPTOALTSTACK, OP_DUPFROMALTSTACK: copy to / from alt stack without 
removing the item

14. UTXO commitment: good but not in near future

15. Starting as a softfork, moving to a hardfork? SW Softfork is a quick 
but dirty solution. I believe a hardfork is unavoidable in the future, 
as the 1MB limit has to be increased someday. If we could plan it ahead, 
we could have a much cleaner SW hardfork in the future, with codes 
pre-announced for 2 years.


[1] https://bitcointalk.org/index.php?topic=181734.0
[2] 
https://lists.linuxfoundation.org/pipermail/bitcoin-dev/2015-November/011679.html
[3] https://blockstream.com/2015/08/24/treesignatures/
[4] https://bitcointalk.org/index.php?topic=1106586.0
[5] 
https://lists.linuxfoundation.org/pipermail/bitcoin-dev/2015-September/010977.html
[6] https://bitcointalk.org/index.php?topic=58579.msg690093#msg690093