fix vm roll #451
fix vm roll #451
Conversation
Signed-off-by: Alexandru Gheorghe <alexandru.gheorghe@parity.io>
Signed-off-by: Alexandru Gheorghe <alexandru.gheorghe@parity.io>
|
|
||
| revive_cheat_test!(test_custom_nonce, "Nonce"); | ||
| revive_cheat_test_original!(test_nonce, "Nonce"); | ||
| revive_cheat_test_original!(test_roll, "Roll"); |
There was a problem hiding this comment.
Do not we need to call revive_cheat_test!(test_custom_roll, "Roll")
There was a problem hiding this comment.
The only custom stuff is this: https://github.com/paritytech/foundry-polkadot/pull/451/files#diff-91c2bf2016961cb8b7dadc4b761d94c6529077494b377ddddda2aaab10d07f3fR36, which copied from your PR, but I think is not needed because this test are already running with --polkadot.
There was a problem hiding this comment.
My point was that revive_cheat_test_original is looking for tests in original cheats directory. and this test we have locally in revive dir
| // Set block number in pallet-revive runtime. | ||
| self.0.lock().unwrap().execute_with(|| { | ||
| System::set_block_number(new_height.try_into().expect("Block number exceeds u64")); | ||
| let new_block_number: u64 = new_height.try_into().expect("Block number exceeds u32"); |
There was a problem hiding this comment.
should not we use System::run_to_block?
There was a problem hiding this comment.
The problem with run_to_block or anything that populates all the blocks, is that roll parameter seems to be part of the fuzz test, so that test takes a really long time, I gave up after 10s of minutes.
There was a problem hiding this comment.
OK, should the block hashes be in sync between EVM and revive?
There was a problem hiding this comment.
I see that foundry just do:
#[inline]
fn block_hash_ref(&self, number: u64) -> Result<B256, Self::Error> {
Ok(keccak256(number.to_string().as_bytes()))
}
if they do not have already this in cache.
So it means that foundry does not build blocks to calculate hash, and this is not done in roll. It is generated when we are accessing blockhash as I understand.
New blocks needs to be finalised and initialised otherwise
blockhashopcode won't find the hash and fail.