Well-balanced tree on empty list input

[emielsebastiaan]

We have found an inconsistency between the testnet data by Jam-Duna and our implementation [{PyJAMaz}].
This is related to the implementation of GP-0.6.2-eq:7.3 and especially variable r.

Difference

"jam-duna_value": "0xbc36789e7a1e281436464229828f817d6612f7b477d66591ff96a9e064bcc98a"
"pyjamaz_value": "0x0000000000000000000000000000000000000000000000000000000000000000",

Suspected causes

JamDuna creates a Keccak-hash over payload '0x00', while we suspect an empty list should be input into the well-balanced tree (GP-0.6.2-eq:E.2), which in turn calls function N (GP-0.6.2-eq:E.1). For an empty list (|v|=0) the zero-hash should be the output.

References

• JAM Dune Testnet Data:

  jamtestnet/data/fallback/state_snapshots/1_000.json

  Line 195 in 0b5b2d3

  "0xbc36789e7a1e281436464229828f817d6612f7b477d66591ff96a9e064bcc98a"

• GP-0.6.2-eq:7.3: https://graypaper.fluffylabs.dev/#/5f542d7/0fec010fec01
• GP-0.6.2-eq:E.1: https://graypaper.fluffylabs.dev/#/5f542d7/394701394701
• GP-0.6.2-eq:E.2: https://graypaper.fluffylabs.dev/#/5f542d7/3a14003a1400

[sourabhniyogi]

Thank you, will study! ==> we have studied and agree with you, will publish soon!

[charliewinston14]

Reported the same issue here #36

[sourabhniyogi]

@emielsebastiaan Can you kindly check this issue is still present in this recent commit? Thank you!

[sourabhniyogi]

Closing this based on @charliewinston14 comment here -- thank you for the detailed report!