dakk / bitcoinml Goto Github PK

https://testnet.smartbit.com.au/block/54504

lian/bitcoin-ruby@8ae149a

too many opcodes in script

Instead of integrating huge external deps, we create a signature engine abstraction; so:

• bitcoinml alone can't verify scripts
• a bitcoinml user can provide his custom Signature implementation (ie wrapping secp256k1), if needs script verification

Introduce an optional argument to parse and serialize function called "hex" for handling hex data.

• Tx
• Tx tests
• Block
• Block tests

The block after: https://www.blocktrail.com/tBTC/block/000000000207e63e68f2a7a4c067135883d726fd65e3620142fb9bdf50cce1f6 fails the check_target

Travis-ci could reduce the number of error I made publishing new releases on opam-repository.

Provide a module to compute a tx list merke root:

module Merkle : sig
	type t
	val of_txs : Tx.t list -> t
end

Integrate bitcoin network protocol message parse and serialize.

• Move message.ml* from caravand to proto.ml* in bitcoinml
• Docs
• Integrate tests

Many times it's not necessary to parse the full block but only the headers; so we could define a new type:

type Block.l = {
   header: Header.t;
   txs: Tx.t list Lazy.t;
}

In case of 2x fork, we need to add a new network for 2x, and update checkpoint list for both chains (just for redundancy, letchain doesn't need that)

Provide the ability to create a raw transaction, sign it with a keypair, and create the complete tx.

The upcoming activation of segwit, push the need of segwit transaction format support

https://bitcoincore.org/en/segwit_wallet_dev/#transaction-serialization
https://bitcoin.stackexchange.com/questions/49097/what-does-a-segregated-witness-transaction-look-like

The Tx could be a structure like now, but with optional witness fields:

type Witness.t = {
	marker		: uint32;
	flag			: uint32;
	witness		: field list;
	size		: int;
}

type t = {
	hash			: Hash.t;
	version		: int32;
	txin 			: In.t list;
	txout 		: Out.t list;
	locktime		: uint32;

	witness		: Witness.t option;
	size			: int;
} [@@deriving sexp]

The witnessScript reside in a different parte of the tx, but every witnessScript is associated to one txin; could be a smart trick to inflate the wintnessScript inside the txin.

Use this: http://n.bitcoin.ninja/checktx for examples and tests

Tasks:

• Witness abstract data-structure
• Parsing
• Serialization
• Tx hash and tx size for segwit
• Tx vsize
• Witness tests
• Witness tests - P2WPKH
• Witness tests - P2WSH
• Witness tests - P2SH(WPKH)
• Witness tests - P2SH(WSH)
• Witness tests - Hash and sizes
• Block hash and block size for segwit

for:
020000000001018123858b74cc156be287dbac5442508483bfe6d9a4165712e5bd0b65a9120a210100000000ffffffff0250c30000000000002200207110c7d8cf7bfa8c93d190d699571c128eac6e922ce0ca3bdf0bc3495109dffc14d606000000000016001499b9280ca4f5eeea13e51770e7a30eb512ed473702483045022100952394943abe1e75d20d4b20622416d0618585bb3266a8122894024a7cfcc6c8022051465780d56717548951fad188fe038f8820c1e054d3a9b90533084cb06625cf012102a3a3756f1e06a4b4764cc3fcf2e804257727e4431b548741f8be057f1628e39900000000

d80c367ee53355ff52bb818545e73f806965a056d2085771e5c3c994af9f640a
but library returns:
df986955f737c0d90fbedc85580bc21ec3951b12dc68b8c611e48c9d4386bb74

Create a module for address representation

Modify the script.to_string to propose an hex encoding for script data

Bitcoinml doesn't provide parameters for different blockchains but it could be useful for many use cases; could be useful to move them from https://github.com/dakk/letchain/blob/master/src/blockchain/params.ml to bitcoinml.

Even in the test suite we use part of them (for address decoding)

• Modify Script apis to allow passing of segwit stack data
• Enable spendable_by for witness addresses
• Provide a Script.verify working implementation
• Primitives for creating common scripts
• Testing framework for script verification
• Integrate secp verifier on test suite

The interested block is the block next to:

54506 : 0000000001df68704402058e0947804685f6a647ff1eeee6ff038a1f722c4676

There are about 1MB of INVALID_OPCODEs.

This bug stops letchain from syncing testnet.

• Merkle tree root real bitcoin test
• Block.Header.check_target tests
• Segwit transaction tests
• Multi-input witness txs
• Lazy block parsing

For better handling script templates, we create template modules for each standard script template:

• Pubkey - input
• Pubkey - output
• Pubkey - test
• Pubkeyhash - input
• Pubkeyhash - output
• Pubkeyhash - test
• Scripthash - input
• Scripthash - output
• Scripthash - test
• Multisig - input
• Multisig - output
• Multisig - test
• Witnessscripthash - input
• Witnessscripthash - output
• Witnessscripthash - test
• Witnesspubkeyhash - input
• Witnesspubkeyhash - output
• Witnesspubkeyhash - test
• WItnesscommitment - output
• Witnessscripthash - test

• address.mli
• base58.mli
• block_lazy.mli
• block.mli
• hash.mli
• merkle.mli
• params.mli
• script_verify.mli
• script.mli
• tx.mli
• varint.mli
• main

The addresses are always converted with mainnet prefixes; we need to extends Tx.Out.spendable_by from bitcoinml with a prefix optional argument.

Integrate coin params for networks, and implement moneyrange checks like https://github.com/bitcoin/bitcoin/blob/master/src/amount.h