question about "hash(false)" when "chainId" > 0

[warren-bank]

hello. please correct me if I'm wrong, but my understanding of EIP 155 is that the raw data Array that gets hashed should contain the value 0 for the fields: r and s.

in looking at your implementation of the hash() function, it also appears that's what is intended.

however, in testing the result.. I'm seeing that these fields are being set to empty (unallocated, zero-length) Buffers.

here's a demo (based mostly on the example in the README)..

const EthereumTx = require('ethereumjs-tx')
const privateKey = Buffer.from('e331b6d69882b4cb4ea581d88e0b604039a3de5967688d3dcffdd2270c0fd109', 'hex')

const txParams = {
  nonce: '0x00',
  gasPrice: '0x09184e72a000', 
  gasLimit: '0x2710',
  to: '0x0000000000000000000000000000000000000000', 
  value: '0x00', 
  data: '0x7f7465737432000000000000000000000000000000000000000000000000000000600057',
  // EIP 155 chainId - mainnet: 1, ropsten: 3
  chainId: 3
}

const tx = new EthereumTx(txParams)
tx.sign(privateKey)

let raw = tx.raw
tx.hash(false)  // bit of a hack: "raw" contains updated {v,r,s} from "items"

console.log(raw.slice(6))  // print the last 3 fields: [v,r,s]

output:

[ <Buffer 03>, <Buffer >, <Buffer > ]

What's really odd (to me) is that both of these fields are marked "allowZero". So, without tracing the code in much depth, it would seem that the setter for these fields should allow the assignment to 0.

Is this a bug?

[warren-bank]

quick bit of tracing later..

the setter calls ethereumjs-util.toBuffer(0).. which returns ethjs-util.intToBuffer(0).. which return new Buffer('00', 'hex').. which looks good

in the setter itself, there's some odd logic going on:
if (field.allowLess && field.length) v = exports.stripZeros(v)
..which strips the 0 out of the Buffer
..leaving it empty (zero-length)

fwiw, you might want to compare the logic in this setter to here

[fanatid]

When checked code, I asked myself about same question. Also if chainId is not zero and tx already have signatures they will be lost. Ping @holgerd77

[davidmurdoch]

Related to the behavior of utils.stripZero and this issue: https://github.com/ethereumjs/ethereumjs-util/issues/141

[danjm]

This will likely be resolved when we resolve https://github.com/ethereumjs/ethereumjs-tx/issues/112

[ryanio]

This should now be resolved with #812. Please use the new tx.getMessageToSign() in place of tx.hash(false).