fix(monad): exclude blob suites from MONAD_EIGHT

[haythemsellami]

Summary

• add a blob-fork validity marker and apply it to the EIP-4844 blob suites so they do not collect for blobless forks like MONAD_EIGHT
• make MONAD_EIGHT stop advertising blob transaction support while keeping the rest of the fork behavior intact
• carry sender authority history through fill so the forks/monad_eight branch can fill state fixtures again

Verification

• UV_CACHE_DIR=/tmp/uv-cache uv run --no-sync pytest packages/testing/src/execution_testing/cli/pytest_commands/plugins/forks/tests/test_markers.py -q
• UV_CACHE_DIR=/tmp/uv-cache uv run --no-sync pytest packages/testing/src/execution_testing/forks/tests/test_forks.py -q -k monad_eight_disables_blob_support
• UV_CACHE_DIR=/tmp/uv-cache uv run --no-sync fill --fork MONAD_EIGHT --clean --no-html --single-fixture-per-file --output /tmp/monad-fill-berlin-sample -m state_test tests/berlin/eip2929_gas_cost_increases/test_call.py -k test_call_insufficient_balance -q
• UV_CACHE_DIR=/tmp/uv-cache uv run --no-sync fill --fork MONAD_EIGHT --clean --no-html --single-fixture-per-file --output /tmp/monad-fill-blob-sample -m state_test tests/cancun/eip4844_blobs/test_excess_blob_gas.py -q # no tests ran

[greptile-apps]

Greptile Summary

This PR introduces a valid_for_blob_forks pytest marker to exclude blob-specific EIP-4844 test suites from blobless forks like MONAD_EIGHT, adds comprehensive blob-related method overrides to the MONAD_EIGHT fork class so it no longer advertises blob transaction support, and fixes a senders_authorities propagation gap in make_hive_fixture that was breaking state-fixture fills on the forks/monad_eight branch.

Key changes:

• New ValidForBlobForks flag-type validity marker in forks.py that, when applied, subtracts all blobless forks (and their transition forks) from the test's collected fork set.
• MONAD_EIGHT now overrides eight blob-related class methods to return False/None, while retaining the KZG point-evaluation precompile (address 0x0a).
• senders_authorities dict is now properly initialised and threaded through each generate_block_data call in make_hive_fixture, matching the existing behaviour in make_fixture.
• The valid_for_blob_forks() marker is applied to all seven EIP-4844 blob test modules.

Coverage concern — KZG precompile tests incorrectly excluded:
The valid_for_blob_forks() marker has been applied to test_point_evaluation_precompile.py and test_point_evaluation_precompile_gas.py, which will suppress them entirely when running against MONAD_EIGHT. However, these tests only issue regular (non-blob) transactions that call the KZG precompile directly — they do not require blob transaction support. Because MONAD_EIGHT explicitly retains the point-evaluation precompile, this coverage gap is unintentional and should be reviewed.

Confidence Score: 2/5

• Core logic is sound but has an unintended test coverage gap: KZG precompile tests are excluded from MONAD_EIGHT despite that fork supporting the precompile.
• The marker mechanism, MONAD_EIGHT fork overrides, and senders_authorities propagation are all correctly implemented and well-tested. However, applying valid_for_blob_forks() to the KZG precompile test files is too broad — these tests don't require blob transactions, only the KZG precompile which MONAD_EIGHT explicitly retains. This creates a silent test coverage loss that directly contradicts the stated goal of keeping the precompile functional on MONAD_EIGHT.
• tests/cancun/eip4844_blobs/test_point_evaluation_precompile.py (line 63) and tests/cancun/eip4844_blobs/test_point_evaluation_precompile_gas.py (line 31) — consider removing valid_for_blob_forks() or introducing a narrower marker that distinguishes blob transaction requirements from blob infrastructure support.

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A[pytest collects test] --> B{has valid_for_blob_forks marker?}
    B -- No --> C[ValidForBlobForks returns empty set\nforks - empty = unchanged fork set]
    B -- Yes --> D[ValidForBlobForks._process_with_marker_args\nbuilds set of blobless forks]
    D --> E[fork.supports_blobs == False?]
    E -- Yes --> F[Add fork + its transition forks to exclusion set]
    E -- No --> G[Skip fork]
    F --> H[process: forks - exclusion set]
    G --> H
    C --> I[Test runs for all selected forks]
    H --> I

    subgraph MONAD_EIGHT overrides
        J[supports_blobs → False]
        K[header_excess_blob_gas_required → False]
        L[header_blob_gas_used_required → False]
        M[tx_types → Prague types minus type-3]
        N[blob_schedule → None]
        O[engine_get_blobs_version → None]
        P[engine_new_payload_blob_hashes → False]
        Q[full_blob_tx_wrapper_version → None]
    end

    D -->|checks| J

Loading

_{Last reviewed commit: 6c37550}

[pdobacz]

Nice, thank you so much for this!

Can you switch the target to the default forks/monad_nine. I believe some of the diff should disappear.

Also, does this setup now allow to fill cancun/eip4844... tests with some meaningful result? if so, we may want to un-exclude them in the .github/configs/feature.yaml fill commands.

[haythemsellami]

Nice, thank you so much for this!

Can you switch the target to the default forks/monad_nine. I believe some of the diff should disappear.

Also, does this setup now allow to fill cancun/eip4844... tests with some meaningful result? if so, we may want to un-exclude them in the .github/configs/feature.yaml fill commands.

@pdobacz I addressed all the issues/bot's comments.

For cancun/eip4844..., I don’t think we should un-exclude the whole group in .github/configs/feature.yaml yet. The actual blob-tx/blob-header suites are still not meaningful for Monad because MONAD_EIGHT/MONAD_NINE do not support blobs. The point-evaluation precompile suites are meaningful now, but they currently fail on both MONAD_EIGHT and MONAD_NINE with real behavior mismatches, so I’d keep the broad exclusion for now and narrow it only after we fix those KZG failures.