CIP-36 test vector CBOR incorrect

[rooooooooob]

as per schema.cddl:

delegation = ($voting_pub_key, $weight)

key_registration = {
  1 : [+delegation] / legacy_key_registration,
  2 : $stake_credential,
  3 : $reward_address,
  4 : $nonce,
  ? 5 : $voting_purpose .default 0
}

you can see the non-legacy delegation (key 1) is an array of delegation, a basic group that isn't array-encoded where it's defined.

However the test vector for the non-legacy format is a119ef64a50182825820a6a3c0447aeb9cc54cf6422ba32b294e5e1c3ef6d782f2acff4a70694c4d16630182582000588e8e1d18cba576a4d35758069fe94e53f638b6faf7c07b8abd2bc5c5cdee0302582086870efc99c453a873a16492ce87738ec79a0ebd064379a62e2c9cf4e119219e03581de0ae3a0a7aeda4aea522e74e4fe36759fca80789a613a58a4364f6ecef041904d20500 which decoded (using cbor.me) is:

{
    61284: {
        1: [
              [h'A6A3C0447AEB9CC54CF6422BA32B294E5E1C3EF6D782F2ACFF4A70694C4D1663', 1],
              [h'00588E8E1D18CBA576A4D35758069FE94E53F638B6FAF7C07B8ABD2BC5C5CDEE', 3]
        ],
        2: h'86870EFC99C453A873A16492CE87738EC79A0EBD064379A62E2C9CF4E119219E',
        3: h'E0AE3A0A7AEDA4AEA522E74E4FE36759FCA80789A613A58A4364F6ECEF',
        4: 1234,
        5: 0
    }
}

likewise the final human-readable part of the test vector says (2 element array with 2 element arrays as the elements):

{
  61284: {
    1: [['0xa6a3c0447aeb9cc54cf6422ba32b294e5e1c3ef6d782f2acff4a70694c4d1663', 1], ['0x00588e8e1d18cba576a4d35758069fe94e53f638b6faf7c07b8abd2bc5c5cdee', 3]],
    2: '0x1c5d88aa573da97e5a4667e0f7c4a9c6a3d848934c3b0a5b9296b401540f2aef',
    3: '0xe0ae3a0a7aeda4aea522e74e4fe36759fca80789a613a58a4364f6ecef',
    4: 1234,
    5: 0
  },
  61285: {
    1: '0x3aaa2e6b43c0a96e880a7d70df84dffb2a1a17b19d7a99a6ed27b91d499b32027c43acfbf6dff097af7634b2ee38c8039af259b0b6a64316f02b4ffee28a0608'
  }
}

According to the schema.cddl spec that key should actually be (one 4-element array):

        1: [
              h'A6A3C0447AEB9CC54CF6422BA32B294E5E1C3EF6D782F2ACFF4A70694C4D1663',
              1,
              h'00588E8E1D18CBA576A4D35758069FE94E53F638B6FAF7C07B8ABD2BC5C5CDEE',
              3,
        ],

The test vectors seem to assume that it was instead defined as delegation = [$voting_pub_key, $weight].

[rooooooooob]

Kind of related, but not as big of a deal, is that the CBOR there also has a different stake credential between the human readable final output (0x1c5d88aa573da97e5a4667e0f7c4a9c6a3d848934c3b0a5b9296b401540f2aef) and the ones beforehand (0x86870efc99c453a873a16492ce87738ec79a0ebd064379a62e2c9cf4e119219e). This is the case for both the legacy and new test vectors.

The legacy delegation also differs between the human readable and CBOR parts. In the CBOR it's the same as the 1st one in the non-legacy delegation array (0xa6a3c0447aeb9cc54cf6422ba32b294e5e1c3ef6d782f2acff4a70694c4d1663) instead of the 0x0036ef3e1f0d3f5989e2d155ea54bdb2a72c4c456ccb959af4c94868f473f5a0 in the human-readable parts (both intermediate and final).

[rphair]

@rooooooooob @Ryun1 perhaps while CIP-0036 is under scrutiny from #373 we could have a corrected CBOR submitted as a PR? I'll watch out for this & get it classified as a "correction" immediately so we can put the change through quickly.

[Ryun1]

@rphair

Yes, this is a good spot from @rooooooooob, this is a mistake in the CDDL and keys used to generate test vectors - I have checked this with Catalyst and one of the authors; @kevinhammond.

Tomorrow I will create a PR with the correction. Thanks for the heads up.

[Ryun1]

@rphair @rooooooooob

Please see PR with fixes: #426

[rphair]

thanks @rooooooooob @Ryun1 @kevinhammond ... closing this so further review can continue in #426.