The address range used by rollups is between 0x000 and 0xFFF? I would love for rollups to get use from the tests here.

I think we could optionally add something into to the framework, where addresses that dont fit within that range are bumped. Although that might be tricky.

We could add an error check if a test uses an address that fits within that range nonetheless.

Hi @spencer-tb, as per https://eips.ethereum.org/EIPS/eip-7587

The address range between 0x0000000000000000000000000000000000000100 and 0x00000000000000000000000000000000000001ff is reserved for use by the RIP process.

At Kakarot we are extensively using ethereum/tests for our zkevm, and we recently implement RIP7212, a secp256r1 signature verification available at address 0x100. This caused conflicts with some existing tests, so I bumped all the addresses in the range 0x100-0x1ff and re-filled the fixtures, so that we can keep our test coverage - and it worked very well!

I also refilled the legacy (non-python) tests, you can check here for more info:
https://github.com/kkrt-labs/tests

Hi, I think #584 is the beginning to make a solid foundation to prevent this from ever being an issue again.

All tests written with the new convention should start from 0x1000 but, adding to that, this start address is now even configurable so rollups could do:

fill -n auto --test-contract-start-address 0x1000000000000000000 --test-contract-address-increments 0x1000000000000000000

to specify from which address the test contracts should be defined.

The PR does not update all contract addresses, but we will start the refactor effort for the rest of the tests in the following weeks.

Hi !
Indeed this was just a quick fix, a more robust and customizable solution is definitely the right approach - especially now that rollups will start innovating and use different specific ranges of reserved addresses.

This is just a "quick fix", let me know if you want this to get merged or simply wait for the full refactor.