New witness program definition in BIP141, and related revision in 142 - 144

[jl2012]

BIP141: New witness program definition
BIP142: Title change; new address definition
BIP143: Title change and update reference implementation
BIP144: Update reference implementation

[CodeShark]

ACK on the 141 and 143 changes

[CodeShark]

ACK on all BIPs.

[dcousens]

ACK BIP 142, much improved.

[schildbach]

signle > single

[jl2012]

@schildbach Corrected. Thanks

[btcdrak]

ACK

[NicolasDorier]

NACK for 142. This would make a conflict between two version bytes, breaking client who depended on the base58 to infer the network of a base58 string.

Main EXT_PUBLIC_KEY 4,136,178,30 <=> TestNet WITNESS_P2WPKH 4

Solution without conflict:

base58Prefixes[(int)Base58Type.WITNESS_P2WPKH] = new byte[] { (0x03) };
base58Prefixes[(int)Base58Type.WITNESS_P2WSH] = new byte[] { (40) };

base58Prefixes[(int)Base58Type.WITNESS_P2WPKH] = new byte[] { 0x6 };
base58Prefixes[(int)Base58Type.WITNESS_P2WSH] = new byte[] { (10) };

I also want to note there is conflict with the segnet and mainnet as well

Main SCRIPT_ADDRESS 5 <=> segnet EXT_PUBLIC_KEY 5,53,135,207
Main SCRIPT_ADDRESS 5 <=> segnet EXT_SECRET_KEY 5,53,131,148

@sipa I think this should be changed for the public segnet, or people trying the segnet will break when they wanted to extract the network from the base58 string.

It's a weak NACK, I understand that not making collision will be harder and harder though.

[jl2012]

BIP142 revised to resolve the conflict

[NicolasDorier]

Thanks, last thing I am thinking. Why making a scheme address specific for v0 and not something general to all segwit payment ? If we use the BIP142 scheme, the problem is that when a new segwit will release, no wallet will be able to send to it... all wallets want to do is sending money, they should not need to implement a new address scheme for a new segwit version.

There should be one address version for both wsh and wpkh.

[jl2012]

@NicolasDorier This is actually generalizable, as long as the future version is 20 byte or 32 byte or both. It's very likely that future versions will be 32 byte.

In the very first version of BIP142 I actually made it completely generalizable to any script length but that idea was not very popular: #267

[NicolasDorier]

well, in future version a new address version should be found. And during a while, all wallet will not be able to send to it.

[NicolasDorier]

I see, ACK. In the future when a new segwit version will be out, maybe a new address scheme will be implemented which is generaly secure for scripts.

[luke-jr]

Don't need a BOM here...