test: validate JSON-RPC schema against execution-api fixtures

[XinyuCRO]

Dependent PR:

• feat(rpc): add eth_baseFee JSON RPC method #1027
• fix(rpc): reject block numbers without hex prefix #1028
• fix(evm): allow access list creation without requiring gas when no authorization list is provided. #1029
• fix(rpc): align eth_getStorageAt storage key parsing #1030
• feat(rpc): include txHash in debug_traceBlock JSON RPC response #1031
• feat(rpc): default Block param to latest on state methods. #1033

doc update PR:

• doc: update RPC schema differences cronos-docs#338

Tests:

• Add JSON-RPC schema tests against upstream execution-apis fixtures
• Sync fixtures from upstream ethereum/execution-apis before running the schema tests

[codecov]

Codecov Report

❌ Patch coverage is 51.51515% with 16 lines in your changes missing coverage. Please review.
✅ Project coverage is 46.21%. Comparing base (099ce78) to head (6d1d30c).

Files with missing lines	Patch %	Lines
rpc/namespaces/ethereum/eth/api.go	0.00%	8 Missing ⚠️
x/evm/keeper/grpc_query.go	12.50%	6 Missing and 1 partial ⚠️
rpc/types/block.go	0.00%	0 Missing and 1 partial ⚠️

Additional details and impacted files

@@             Coverage Diff             @@
##           develop    #1010      +/-   ##
===========================================
+ Coverage    46.18%   46.21%   +0.02%     
===========================================
  Files          197      197              
  Lines        17214    17237      +23     
===========================================
+ Hits          7951     7966      +15     
- Misses        8433     8443      +10     
+ Partials       830      828       -2     

Files with missing lines	Coverage Δ
rpc/backend/account_info.go	83.72% <100.00%> (+2.14%)	⬆️
x/evm/types/tracer.go	35.08% <ø> (ø)
rpc/types/block.go	53.93% <0.00%> (+1.18%)	⬆️
x/evm/keeper/grpc_query.go	71.66% <12.50%> (-0.24%)	⬇️
rpc/namespaces/ethereum/eth/api.go	0.00% <0.00%> (ø)

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[JayT106]

@claude

[github-actions]

Claude finished @JayT106's task in 3m 26s —— View job

---

PR Review — test: validate JSON-RPC schema against execution-api fixtures

• Gather context and understand the PR
• Read CLAUDE.md (not present)
• Review changed Go files (rpc/, x/evm/)
• Review new Python test files
• Check for state-breaking changes
• Post inline comments on issues found
• Write final review summary

---

Summary

This PR adds:

1. A new eth_baseFee RPC method
2. A fix for eth_getStorageAt key parsing
3. Stricter block number parsing (hex-only)
4. txHash in debug_traceBlock results
5. A fix to getAccessListExcludes guard condition
6. A comprehensive JSON-RPC schema test suite against upstream execution-apis fixtures

Overall the code quality is good and the test infrastructure is well-designed. See findings below.

---

Findings

🔴 Breaking Change — Stricter Block Number Parsing (rpc/types/block.go)

The removal of the decimal fallback is a client-facing breaking change. Before, BlockNumber.UnmarshalJSON accepted decimal strings like "12345" by falling back to cast.ToUint64. Now any non-hex input (missing 0x prefix) returns an error.

Any client relying on decimal block number inputs will break silently. The PR description calls it "stricter" but doesn't highlight this as a compatibility concern. This is acceptable if there are no known users of that path, but should be explicitly acknowledged.

🟡 Unused Return Value in decodeStorageKey (rpc/backend/account_info.go:140)

storageKey, _, err := decodeStorageKey(key)

The inputLength int return value is discarded everywhere it's called. If it was intended for future use (e.g., detecting right-vs-left padding for legacy compat), it should be documented; otherwise simplify the signature.

🟡 No Unit Test for eth_baseFee (rpc/namespaces/ethereum/eth/api.go)

The BaseFee() method has 0% coverage per Codecov. This method has two code paths (head.BaseFee == nil and non-nil) — a unit test would be straightforward given the existing backend.go interface.

🟡 getAccessListExcludes logic change (x/evm/keeper/grpc_query.go:1052)

The original code checked args.Gas == nil unconditionally when AuthorizationList was set. The new code wraps it in if len(args.AuthorizationList) > 0, which is correct — but note this means that if a caller passes authorizationList: [] (empty slice), gas is no longer checked. This is intentional per the fix, but worth confirming the behavior is consistent with go-ethereum.

🟡 CodeQL: Implicit string concatenation (_schema_constants.py:162-165 and :166-168)

LEGACY_CREATE_BYTECODE = (
    "0x600d380380600d6000396000f360004381526020014681526020014181526020014"
    "881526020014481526020013281526020013481526020016000f3"
)

Python implicitly concatenates adjacent string literals. These aren't inside a list so there's no actual bug, but using an explicit + or a single long string removes the CodeQL noise and makes intent clearer.

🟡 CodeQL: Bare except Exception: pass without comment (conftest.py:117)

The _w3_wait_for_block function silently swallows all exceptions in the polling loop. Line 207 has a good comment explaining why ("The RPC endpoint can reject early requests..."); the identical pattern at line 117 does not. A brief comment here would satisfy CodeQL and aid future readers.

🟢 TraceBlock txHash addition (x/evm/keeper/grpc_query.go:626)

This is a clean, minimal fix. TxTraceResult.TxHash was already defined in types/tracer.go but never populated. The addition is correct and non-breaking.

🟢 decodeStorageKey implementation

Leniently accepts keys with or without 0x prefix and handles odd-length hex. common.BytesToHash right-aligns the bytes as expected for storage slot addressing. The error paths return a structured InvalidParamsError with the offending key in the message. Well done.

🟢 eth_baseFee implementation

Returns new(big.Int) (zero) when head.BaseFee == nil (pre-EIP-1559 contexts) rather than nil, which avoids a null pointer in the JSON response. Correct behavior.

🟢 Python Test Infrastructure

The fixture sync architecture (_execution_apis_sync.py, genesis overlay injection, conftest.py setup) is well-structured. The separate normalizers, rewriters, constants, and report files follow good separation of concerns. The SCHEMA_MISMATCH_WHITELIST and EXCLUDED_SCHEMA_SPEC_CASES allow-lists are explicitly documented with reasons.

One note: _execution_apis_sync.py fetches from main branch of ethereum/execution-apis at runtime — consider pinning to a specific commit/tag for reproducibility (or exposing a EXECUTION_APIS_REF env var alongside the existing EXECUTION_APIS_ARCHIVE_URL).

---

State-Breaking Changes

None detected. All changes are either:

• Read-only RPC handlers (no state mutations)
• Trace/debug query improvements (no consensus impact)
• Defensive guard clauses in getAccessListExcludes (no state change)

The stricter block number parsing is API-breaking for RPC clients, not a consensus/state-breaking change. It is safe to merge to non-release branches.

---

| Branch: feat/rpc

[JayT106]

in https://github.com/crypto-org-chain/cronos-docs/pull/338/changes#diff-70121135e802c99e062c63c7323c1e41506bb2741106a23b08a05af4a91ad73eR125 mentioned the eth_simulateV1 but this PR seems remove a lot of files, could you give context of it?

[XinyuCRO]

in https://github.com/crypto-org-chain/cronos-docs/pull/338/changes#diff-70121135e802c99e062c63c7323c1e41506bb2741106a23b08a05af4a91ad73eR125 mentioned the eth_simulateV1 but this PR seems remove a lot of files, could you give context of it?

@JayT106 the deleted simulateV1 specs files are copied from execution specs, this pr introduce a way to sync all specs fixtures including simulateV1 ones into the test context without needing to track them in git, should have no behavior changes

[thomas-nguy]

• We probably could split this PR into 2 or 3 differents PRs which clear intent
• Missing changelog

[thomas-nguy]

lgtm