🐞 Sample data are inconsistent

[JohnLCaron]

Is there an existing issue for this?

• I have searched the existing issues

Current Behavior

electionguard/data/0.95.0/sample data

uses base10 or base16 for encoding bigintegers
sometimes use current primes, sometimes previous ones

                             bigint   constants
                             --------  ------------ 
full                         base10   standard
hamilton-general             base16   ?
minimal                      base10  previous
small                        base10  previous

Expected Behavior

No response

Steps To Reproduce

No response

Environment

- OS:

Anything else?

No response

[keithrfung]

That's super interesting. They were all generated with the same codebase and same primes recently. Is this referring to what's include in the zip or what's included in the raw folders?

[keithrfung]

@JohnLCaron can you double check me and look and see how we're serializing in Python to doublecheck there isn't something fishy about the BigInt serialization in terms of base16 vs base10. They should all be using the new primes and base16 aka hex for serialization.

[keithrfung]

We can likely do something post 1.4 where we rebuild the entire set of sample data at that juncture and only include the manifests in here. I think that's the simplest solution.

[JohnLCaron]

ok, the constants are ok now.

however, the coefficients.json files seem wrong, and different than what I get when i run TestEndToEndElection on python library HEAD:

my run

{
  "coefficients": {
    "1": {
      "data": "05"
    },
    "2": {
      "data": "FFE7"
    },
    "3": {
      "data": "0A"
    },
    "4": {
      "data": "FFEC"
    },
    "5": {
      "data": "01"
    }
  }
}

what i see in electionguard/data/0.95.0/sample/hamilton-general/election_record/coefficients.json (actually all the sample files):

{
  "coefficients": [
    {
      "data": "03"
    },
    {
      "data": "FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF40"
    },
    {
      "data": "01"
    }
  ]
}

[JohnLCaron]

Since 0.95 data are now inconsistent with 1.0, Im removing my tests on 0.95 data and closing this issue.