rlp_decoded vs provided rlp inconsistency about execution-spec-tests

[spencer-tb]

This case will be invalid for evmone!

The valid tx rlp is overwritten with the invalid tx rlp in this test: https://github.com/ethereum/execution-spec-tests/blob/main/tests/cancun/eip4844_blobs/test_blob_txs.py#L1155-L1156

I think we can update how the rlp_decoded txs are output in the fixture by deserialzing the invalid tx rlp. Or even simpler for these specific cases we don't display rlp_decoded.

[winsvega]

actually I thought about rlp_decoded.
the only case where it can;t be done is when rlp is invalid. (and this ones are rlp serialization tests)

but we can actually agree to print rlp_decoded to specificly indicate that the block has some field missing.
so if blobGasUsed field is missing it will be like
"blobGasUsed" : "not included in rlp" or smth as a hint.
and if there are extra fields it can print it too
"extraField1" : "..."

and then if we construct invalid block with >256 values we can also indicate it in rlp_decoded
so the user will see what is going on in this rlp.

[winsvega]

and this test should not produce any rlp_decoded or should note in it that the transaction rlp is invalid

Error: Unhandled UpwardsException: Failed to verify rlp vs rlp_decoded in filled block: BlobTransaction::fromRLP(RLP) expected to have exactly 14 elements! (cancun/eip4844_blobs/src/GeneralStateTestsFiller/Pyspecs/cancun/eip4844_blobs/test_blob_txs_full.py::test_reject_valid_full_blob_in_block_rlp[fork_Cancun-blockchain_test-one_full_blob_at_the_end_max_txs], fork: Cancun, block: 1)

[marioevz]

I think we could close this now that #427 is merged or is there anything left to be fixed @winsvega ?

[winsvega]

Yes.

[winsvega]

solved for now by not printing rlp_decoded for some cases