MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII;
 format=flowed
Content-Transfer-Encoding: 7bit
Date: Wed, 09 Dec 2015 09:30:23 -0500
From: jl2012@xbt.hk
To: bitcoin-dev@lists.linuxfoundation.org
Message-ID: <ed988315078ef8c06f4bd71463440e63@xbt.hk>
User-Agent: Roundcube Webmail/1.0.5
Subject: [bitcoin-dev] Impacts of Segregated Witness softfork
Precedence: list

Although the plan is to implement SW with softfork, I think many 
important (but non-consensus critical) components of the network would 
be broken and many things have to be redefined.

1. Definition of "Transaction ID". Currently, "Transaction ID" is simply 
a hash of a tx. With SW, we may need to deal with 2 or 3 IDs for each 
tx. Firstly we have the "backward-compatible txid" (bctxid), which has 
exactly the same meaning of the original txid. We also have a "witness 
ID" (wid), which is the hash of the witness. And finally we may need a 
"global txid" (gtxid), which is a hash of bctxid|wid. A gtxid is needed 
mainly for the relay of txs between full nodes. bctxid and wid are 
consensus critical while gtxid is for relay network only.

2. IBLT / Bitcoin relay network: As the "backward-compatible txid" 
defines only part of a tx, any relay protocols between full nodes have 
to use the "global txid" to identify a tx. Malleability attack targeting 
relay network is still possible as the witness is malleable.

3. getblocktemplete has to be upgraded to deal with witness data and 
witness IDs. (Stratum seems to be not affected? I'm not sure)

4. Protocols relying on the coinbase tx (e.g. P2Pool, merged mining): 
depends on the location of witness commitment, these protocols may be 
broken.

Feel free to correct me and add more to the list.