Complex pipeline: sidecars, dynamic libraries, address prediction, init+runtime bytecode

[GlebRadchenko]

---

name: Pull Request
about: Submit a pull request

Description

Type of Change

• Bug fix (non-breaking change that fixes an issue)
• New feature (non-breaking change that adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• Transpiler fix (changes to venom_generator.py or related transpilation logic)
• Backend fix (changes to vyper/ fork - requires justification)
• Documentation update

Testing

• Tests pass locally (cd foundry && forge test)
• Transpilation works (python3.11 testing/test_framework.py --test-all)
• New tests added (if applicable)

Benchmark Impact

Checklist

• My code follows the project's coding style
• I have performed a self-review
• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation
• My changes generate no new warnings
• Any dependent changes have been merged and published

Related Issues

[chatgpt-codex-connector]

💡 Codex Review

yul2venom/yul2venom.py

Lines 1625 to 1631 in 8979fbb

	# Step 4: Write Output
	if runtime_only:
	# Runtime-only mode: just write runtime bytecode
	if runtime_bytecode:
	with open(output_path, "wb") as f:
	f.write(runtime_bytecode)
	print()

Append sidecars in runtime-only output

When --runtime-only is used, this block writes only runtime_bytecode and never appends sidecar_bytecodes. If the deployed runtime uses dataoffset()/datasize() to reference embedded sidecar bytecode (e.g., CREATE/CREATE2 in runtime), those references will now point past the end of the deployed code and the sidecar deployment will fail. The full bytecode path appends sidecars later, so runtime-only output should mirror that layout or sidecar-enabled contracts will break in bench/repro runs that use runtime-only artifacts.

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[Copilot]

Pull request overview

This PR significantly refactors the Yul→Venom pipeline to handle complex immutable ID mapping, nested sidecars, and category-aware testing/transpilation, while also adding debugging and benchmarking tools and normalizing configs/paths.

Changes:

• Reworked yul2venom.py to derive immutable IDs and deployment ordering directly from Yul, support nested sidecars (including two-pass init/runtime sizing and dataoffset/datasize fixes), and make Yul path handling/configs more robust.
• Enhanced the testing framework with smarter contract discovery and category-aware transpilation (transpile_for_testing), added multi-return and inlining debug tools, and wired these into CI workflows.
• Updated Vyper submodule, optimizer rules, IR conventions (ret/invoke), generator return handling, config JSONs (adding sidecar_nonce_start, order, and normalized deployers), test bytecode paths, and gitignore.

Reviewed changes

Copilot reviewed 74 out of 75 changed files in this pull request and generated no comments.

Show a summary per file

File	Description
yul2venom.py	Core pipeline changes: Yul path handling, immutable ID discovery, sidecar support, dataoffset/datasize layout, config generation updates.
vyper	Submodule bumped to a newer Vyper fork to align with updated IR/backend conventions.
testing/test_framework.py	Adds contract discovery, bytecode-type-aware transpilation (transpile_for_testing), and refines test execution paths.
testing/debug_multi_return.py	New CLI tool to debug multi-return/stack behavior and correlate with Forge tests.
testing/config_benchmark.py	New script to benchmark different inlining configurations against representative contracts.
optimizer/yul_source_optimizer.py	Adds/extends source-level Yul optimization rules for calldata checks, panics, address/type validation, and helpers.
ir/basicblock.py	Adjusts operand ordering semantics in IR printing (no global reversal; special-case invoke).
generator/venom_generator.py	Changes return generation to use temp vars via add and orders ret operands as values-first, PC-last.
foundry/test/StateTest.t.sol	Updates runtime binary paths to ../output/... relative to foundry/.
foundry/test/SolcRefTest.t.sol	Same output path normalization for solc reference runtime.
foundry/test/SolcComparisonTest.t.sol	Normalizes paths for solc vs transpiler comparison binaries.
foundry/test/RuntimeInspectTest.t.sol	Normalizes path to transpiled init+runtime binary.
foundry/test/OpcodeTest.t.sol	Normalizes opcode test/runtime binary paths.
foundry/test/MinimalLoopTest.t.sol	Normalizes path to minimal loop runtime binary.
foundry/test/LoopTraceTest.t.sol	Normalizes path to loop trace init binary.
foundry/test/LoopDebugTest.t.sol	Normalizes path to loop-debug init binary.
foundry/test/FeatureTest.t.sol	Normalizes and clarifies usage of various feature test binaries (init and runtime).
foundry/test/ExtendedFeatureTest.t.sol	Normalizes path to MissingFeatures runtime binary.
foundry/test/DebugLoopCheck.t.sol	Normalizes path to LoopCheckCalldata runtime binary.
foundry/test/ControlFlowTest.t.sol	Normalizes paths for loop-check runtime binaries (including alternate filename).
debug/raw_ir.vnm	Updates stored raw VNM IR for multi-return debugging.
debug/opt_ir.vnm	Updates optimized VNM IR snapshot for multi-return debugging.
debug/config_benchmark.json	Adds stored benchmark results for inlining config comparison.
testing/**/*.json configs (many)	Normalize deployers, add sidecar_nonce_start, order, and id metadata for immutables/sidecars.
configs/repro/LoopSwitchRepro.yul2venom.json	Replaces deleted root config with repro-specific one and normalized deployment data.
configs/init/*.yul2venom.json	Adds sidecar_nonce_start and richer constructor/auto-predicted metadata for init contracts.
configs/bench/*.yul2venom.json	Adds sidecar_nonce_start and, where relevant, immutable ordering metadata for bench contracts.
configs/StorageMemory.yul2venom.json	Removes redundant/old StorageMemory config (StorageMemoryTest config remains).
configs/Modifiers.yul2venom.json	Removes duplicate bench Modifiers config in root; bench one retained under configs/bench.
configs/LoopSwitchRepro.yul2venom.json	Removes root LoopSwitchRepro config in favor of configs/repro version.
configs/ControlFlow.yul2venom.json	Removes root ControlFlow bench config in favor of configs/bench/ControlFlow.
configs/AdvancedFeatures.yul2venom.json	Removes root AdvancedFeatures config in favor of bench config.
.gitignore	Ignores per-config output dirs and VNM intermediates.
.github/workflows/transpile-test.yml	Switches CI to test_framework.py --test-all for category-aware transpile+test.
.github/workflows/benchmark.yml	Reuses transpile_for_testing() logic for benchmark setup when cache miss occurs.

Comments suppressed due to low confidence (6)

yul2venom.py:1

• The fallback path for optimized Yul without @src comments only populates offset_to_id but never offset_to_name, so sorted_offsets and name_to_order remain empty and no constructor/auto-predicted entries get order or updated IDs in that case, contrary to the comment. To make the fallback effective, you should either also populate offset_to_name when using setimm_no_src_pattern (e.g., by deriving names from some other mapping) or guard the subsequent name_to_order and name-based update logic so it does not assume offset_to_name is non-empty.
  yul2venom.py:1
• In transpile_object, failures compiling a sub-object are now logged but no longer cause the function to return an error or raise, so the pipeline can continue and emit bytecode even though some sub-objects failed to transpile. To preserve previous behavior and fail fast on such errors, either re-raise the exception here or return a sentinel error value (e.g., return 1) and have callers check for it before proceeding.
  testing/debug_multi_return.py:1
• If subprocess.run raises (e.g., timeout), the function returns from the except block before restoring the working directory back to PROJECT_ROOT, leaving the process stuck in foundry/. Wrap the os.chdir(PROJECT_ROOT) in a finally block so the working directory is restored regardless of success or failure.
  testing/debug_multi_return.py:1
• This code transpiles with --runtime-only (which typically only produces *_opt_runtime.bin) but then copies from output/MultiReturnTest_opt.bin, which may not exist under runtime-only mode and will cause the cp to fail. Either drop the --runtime-only flag here (so _opt.bin is produced) or change the source of the copy to the runtime-only filename that is actually generated.
  yul2venom.py:1
• discover_sidecar_immutables(raw_yul, name_to_value) scans the entire Yul source with a regex on every sub-object iteration, even though the inputs (raw_yul, name_to_value) are constant for the whole transpilation run. To avoid redundant work, compute the mapping once outside the loop (or cache it) and reuse it per sub-object, optionally filtering per sub-object if needed.
  yul2venom.py:1
• discover_sidecar_immutables(raw_yul, name_to_value) scans the entire Yul source with a regex on every sub-object iteration, even though the inputs (raw_yul, name_to_value) are constant for the whole transpilation run. To avoid redundant work, compute the mapping once outside the loop (or cache it) and reuse it per sub-object, optionally filtering per sub-object if needed.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[github-actions]

🔬 Yul2Venom CI Report

Stage	Status
Tests	✅ passed
Benchmark	✅ success

✅ Test Results

  [core ] TinyCall                       ... ✓ 51 bytes [both] (347ms)
  [core ] TryCatchRepro                  ... ✓ 211 bytes [both] (367ms)
  [core ] VoidCall                       ... ✓ 42 bytes [both] (341ms)
  [init ] InitArrayTest                  ... ✓ 899 bytes [init] (608ms)
  [init ] InitCodeTest                   ... ✓ 147 bytes [init] (369ms)
  [init ] InitComplexTest                ... ✓ 646 bytes [init] (494ms)
  [init ] InitConstructorArgsTest        ... ✓ 485 bytes [init] (448ms)
  [init ] InitImmutableTest              ... ✓ 437 bytes [init] (434ms)
  [init ] InitInheritanceTest            ... ✓ 794 bytes [init] (512ms)
  [init ] InitMultiImmutableTest         ... ✓ 1027 bytes [init] (561ms)
  [init ] InitNewChildTest               ... ✓ 3569 bytes [init] (1302ms)
  [init ] InitPayableTest                ... ✓ 133 bytes [init] (365ms)
  [init ] InitStringTest                 ... ✓ 2323 bytes [init] (1151ms)
  [repro] CallOrderRepro                 ... ✓ 125 bytes [both] (350ms)
  [repro] LoopSwitchRepro                ... ✓ 470 bytes [both] (408ms)
============================================================
Results: 47 passed, 0 failed, 47 total
Results saved to /home/runner/work/yul2venom/yul2venom/debug/transpile_results.json
Running Forge tests...
Test Results: 358 passed, 0 failed

📊 Benchmark Results

Yul2Venom Benchmark Report

Generated: 2026-02-06 18:54:46
Baseline: default_200

---

Summary

• Contracts benchmarked: 8
• Transpilation success: 8/8
• Optimization runs tested: [200]
• Solc modes: ['default', 'via_ir', 'ir_optimized']

Bytecode Size (bytes)

Contract	Transpiled	default_200	via_ir_200	ir_optimized
Arithmetic	1586	1368	1368	1650
ControlFlow	1002	962	962	1043
StateManagement	2234	2686	2686	2981
DataStructures	1392	1576	1576	2031
Functions	2051	2452	2452	3477
Events	735	826	826	1163
Encoding	961	976	976	1264
Edge	2694	3112	3112	3787

Size Delta vs Baseline (default_200)

Contract	Transpiled	via_ir_200	ir_optimized
Arithmetic	+15.9%	+0.0%	+20.6%
ControlFlow	+4.2%	+0.0%	+8.4%
StateManagement	-16.8%	+0.0%	+11.0%
DataStructures	-11.7%	+0.0%	+28.9%
Functions	-16.4%	+0.0%	+41.8%
Events	-11.0%	+0.0%	+40.8%
Encoding	-1.5%	+0.0%	+29.5%
Edge	-13.4%	+0.0%	+21.7%

---

Gas Usage (avg gas per function call)

Arithmetic

Function	Transpiled (aggressive+yul-o2)	Native (Solc default_200)	Delta
and_	186	485	-61.6%
eq	201	287	-30.0%
gt	186	221	-15.8%
gte	189	642	-70.6%
iszero	160	770	-79.2%
lt	195	191	+2.1%
lte	189	796	-76.3%
not_	160	707	-77.4%
or_	201	331	-39.3%
safeAdd	244	799	-69.5%
safeDiv	246	697	-64.7%
safeExp	314	524	-40.1%
safeMod	208	477	-56.4%
safeMul	246	783	-68.6%
safeSub	221	601	-63.2%
sar	166	559	-70.3%
sgt	186	265	-29.8%
shl	209	529	-60.5%
shr	186	463	-59.8%
signExtend16	171	652	-73.8%
signExtend8	193	365	-47.1%
slt	186	353	-47.3%
unsafeAdd	186	309	-39.8%
unsafeDiv	208	653	-68.1%
unsafeExp	243	564	-56.9%
unsafeMod	254	741	-65.7%
unsafeMul	188	421	-55.3%
unsafeSub	186	551	-66.2%
xor_	201	243	-17.3%

ControlFlow

Function	Transpiled (aggressive+yul-o2)	Native (Solc default_200)	Delta
breakLoop	1306	1858	-29.7%
continueLoop	1503	1678	-10.4%
earlyReturn	181	354	-48.9%
ifElse	211	294	-28.2%
loopCount	8120	11415	-28.9%
loopSum	683	706	-3.3%
nestedLoop	1513	1778	-14.9%
ternary	229	347	-34.0%
whileLoop	3421	4872	-29.8%

StateManagement

Function	Transpiled (aggressive+yul-o2)	Native (Solc default_200)	Delta
CONST_HASH	154	986	-84.4%
CONST_VALUE	154	876	-82.4%
balances	2358	2533	-6.9%
getArrayElement	443	547	-19.0%
getArrayLength	2253	2245	+0.4%
getMappingValue	2321	2912	-20.3%
getNestedMap	428	1191	-64.1%
getPackedAB	292	547	-46.6%
getStoredBool	2266	2507	-9.6%
getStoredUint	2281	2802	-18.6%
incrementBalance	20315	20593	-1.3%
memoryAlloc	349	726	-51.9%
memoryCopy	1768	3154	-43.9%
popArray	538	1329	-59.5%
pushArray	42450	42720	-0.6%
setMappingValue	20235	21055	-3.9%
setNestedMap	22379	22721	-1.5%
setPackedAB	22333	22998	-2.9%
setStoredBool	20294	20315	-0.1%
setStoredUint	20132	20179	-0.2%

DataStructures

Function	Transpiled (aggressive+yul-o2)	Native (Solc default_200)	Delta
bytesConcat	572	851	-32.8%
bytesLength	224	385	-41.8%
createArray	1585	2099	-24.5%
createStruct	304	510	-40.4%
dynamicArraySum	618	896	-31.0%
fixedArraySum	712	964	-26.1%
processStruct	236	235	+0.4%
processStructArray	1448	2463	-41.2%

Functions

Function	Transpiled (aggressive+yul-o2)	Native (Solc default_200)	Delta
callInternal	207	528	-60.8%
callSelf	692	1171	-40.9%
callVirtualA	142	594	-76.1%
callVirtualB	142	142	+0.0%
factorial	281	487	-42.3%
fibonacci	275	394	-30.2%
interfaceFunc	147	506	-70.9%
nestedInternal	198	302	-34.4%
returnMultiple	178	736	-75.8%
returnNothing	142	350	-59.4%
returnSingle	142	462	-69.3%
selfAdd	184	286	-35.7%

Events

Function	Transpiled (aggressive+yul-o2)	Native (Solc default_200)	Delta
emitBytes	1852	2068	-10.4%
emitComplex	2920	3059	-4.5%
emitIndexed	1581	1676	-5.7%
emitMultiIndexed	2037	2094	-2.7%
emitMultiple	5583	5564	+0.3%
emitSimple	1168	1273	-8.2%
emitString	1822	1969	-7.5%

Encoding

Function	Transpiled (aggressive+yul-o2)	Native (Solc default_200)	Delta
abiEncode	371	603	-38.5%
abiEncodePacked	394	537	-26.6%
abiEncodeWithSelector	356	557	-36.1%
decodePair	228	393	-42.0%
encodeMultiple	655	1005	-34.8%
encodePackedMixed	433	696	-37.8%
keccak256Encode	300	483	-37.9%
keccak256Hash	462	747	-38.2%
keccak256Packed	300	417	-28.1%

Edge

Function	Transpiled (aggressive+yul-o2)	Native (Solc default_200)	Delta
assertCondition	203	565	-64.1%
checkGas	152	728	-79.1%
fallback	189	712	-73.5%
getBlockInfo	188	175	+7.4%
getMsgInfo	221	643	-65.6%
mayFail	302	799	-62.2%
requireTrue	273	598	-54.3%
requireValue	211	433	-51.3%
tryCall	766	1390	-44.9%

---

Configuration

Setting	Value
Baseline	default_200
Optimization Runs	[200]
Solc Modes	['default', 'via_ir', 'ir_optimized']

Mode Descriptions

• default: solc --optimize --optimizer-runs=N (traditional optimizer)
• via_ir: solc --via-ir --optimize --optimizer-runs=N (Yul-based optimizer)
• ir_optimized: solc --ir-optimized --optimize (outputs optimized Yul IR)
• Transpiled: Yul → Venom IR → EVM via Yul2Venom

Interpretation

• Negative delta = smaller than baseline (better)
• Positive delta = larger than baseline (worse)
• FAILED = compilation/transpilation failed

---

Commit: 22d130702ae33349771fa263ad507ed10512666c • 2026-02-06 18:54 UTC

[Copilot]

Pull request overview

Copilot reviewed 97 out of 99 changed files in this pull request and generated 1 comment.

Comments suppressed due to low confidence (11)

yul2venom.py:1

• Bare except ValueError: without handling. Consider logging the error or adding a comment explaining why ignoring is appropriate. If this is intentional (e.g., for hash parsing fallback), add a comment like # Ignore invalid hash format.
  yul2venom.py:1
• The comment states sidecar_nonce_start is 'ALWAYS 1', but it's added as a configurable field. If there are cases where it might not be 1 (as the config structure suggests), clarify those cases in the comment. Otherwise, consider making it a constant rather than a config field.
  generator/venom_generator.py:1
• Using add x, 0 to force materialization creates unnecessary ADD instructions. While the optimizer may eliminate these, consider using a dedicated 'materialize' helper or checking if the optimizer consistently removes these identity operations to avoid unnecessary bytecode bloat.
  generator/venom_generator.py:1
• Duplicate pattern of using add x, 0 for materialization (also at lines 507-512). Consider extracting to a helper method like _materialize_for_ret(val, name) to reduce code duplication and ensure consistent behavior.
  testing/test_framework.py:1
• Comment states tests need BOTH files, but the transpilation strategy only generates one type per category (init -> --with-init, others -> default). Clarify if 'default' mode generates both files or update the comment to match actual behavior.
  yul2venom.config.yaml:1
• Comment 'Testing fix for parser compatibility' suggests a temporary debugging change. If this is the intended production value, update the comment to explain why aggressive is appropriate. Otherwise, revert to the documented default.
  generator/optimizations.py:1
• Multiple bare except ValueError: clauses (5 instances) that silently ignore errors. Consider logging these failures or adding comments explaining why ignoring is safe (e.g., 'Invalid literal format, skip optimization').
  generator/optimizations.py:1
• Multiple bare except ValueError: clauses (5 instances) that silently ignore errors. Consider logging these failures or adding comments explaining why ignoring is safe (e.g., 'Invalid literal format, skip optimization').
  generator/optimizations.py:1
• Multiple bare except ValueError: clauses (5 instances) that silently ignore errors. Consider logging these failures or adding comments explaining why ignoring is safe (e.g., 'Invalid literal format, skip optimization').
  generator/optimizations.py:1
• Multiple bare except ValueError: clauses (5 instances) that silently ignore errors. Consider logging these failures or adding comments explaining why ignoring is safe (e.g., 'Invalid literal format, skip optimization').
  generator/optimizations.py:1
• Multiple bare except ValueError: clauses (5 instances) that silently ignore errors. Consider logging these failures or adding comments explaining why ignoring is safe (e.g., 'Invalid literal format, skip optimization').

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

Pull request overview

Copilot reviewed 97 out of 99 changed files in this pull request and generated 2 comments.

Comments suppressed due to low confidence (8)

yul2venom.py:1

• The function docstring mentions 'sidecar's immutable IDs' but doesn't explain what a sidecar is in this context. Consider adding a brief definition (e.g., 'contracts deployed from parent constructor') for clarity.
  yul2venom.py:1
• Bare except was correctly replaced with except ValueError, but consider whether other exception types (like AttributeError for missing attributes) should also be caught here.
  generator/venom_generator.py:1
• Using add(x, 0) for all return values forces extra ADD operations that may not be eliminated by optimization. Consider whether this pattern is necessary for literals where stack ordering is already deterministic.
  yul2venom.py:1
• The fallback pattern when @src comments are missing may fail to map variable names correctly. Consider logging a warning when falling back to this pattern, as it loses semantic information about variable names.
  testing/test_framework.py:1
• The bytecode type detection relies on directory name matching ('init' in path.parts). This is fragile if directory structure changes. Consider using an explicit config field or convention (e.g., checking for constructor_args in config).
  generator/venom_generator.py:1
• The 0xFFFF mask for 16-bit extraction is well-documented, but consider adding why this is necessary (prevents memory collision from earlier operations). The comment on line 1192-1193 covers this, but it could be more prominent.
  yul2venom.py:1
• Excellent documentation of the bytecode layout calculation. No changes needed.
  optimizer/yul_source_optimizer.py:1
• The improved brace matching using _find_matching_brace is a significant improvement over the previous regex. Well done on handling nested braces correctly.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

The config search priority (package dir first, then CWD upwards) is well-designed. Consider documenting this priority in the docstring for clarity.

[Copilot]

The addition of break_sources to the loop stack tuple is necessary for proper SSA construction, but increases the tuple size from 5 to 6 elements. Consider using a named tuple or dataclass for better maintainability.

[GlebRadchenko]

@codex review

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 3625b3ee22

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Ensure _deployed subobjects compile before init

Sorting subobjects by name ascending does not guarantee _deployed comes before the init object when they share the same depth (e.g., Sidecar sorts before Sidecar_deployed). In that case the init sidecar is compiled first, the runtime entry is missing in data_map, and the code path falls back to the “No runtime found” branch, which produces incorrect datasize/dataoffset for sidecars. This breaks constructors that embed sidecars and rely on datasize("X_deployed")/dataoffset("X_deployed") in the init code. A custom sort (e.g., prioritizing names ending with _deployed) is needed to preserve the intended dependency order.

Useful? React with 👍 / 👎.

[Copilot]

Pull request overview

Copilot reviewed 103 out of 105 changed files in this pull request and generated no new comments.

Comments suppressed due to low confidence (9)

yul2venom.py:1

• The example command references 'MyContract' but the prepare command doesn't specify the output path or config location. Consider adding the full command sequence including the -c flag for the config path to make this example more actionable.
  yul2venom.py:1
• The message references a --force flag that doesn't appear to be defined in the argument parser. Either add the flag or update the message to reflect the actual regeneration mechanism.
  generator/venom_generator.py:1
• Using add x, 0 as a stack ordering forcing mechanism is fragile. This creates unnecessary instructions that depend on the optimizer eliminating them. Consider documenting why this specific pattern is required for stack ordering or implementing a more explicit ordering mechanism.
  generator/venom_generator.py:1
• The mask value 0xFFFF (16 bits) should be defined as a named constant to clarify that this is masking a 2-byte label value. This makes the relationship to the copy_size = IRLiteral(2) above more explicit.
  yul2venom.py:1
• Debug print statements should be removed or placed behind a debug flag before merging to production. These create noise in normal operation.
  testing/test_framework.py:1
• This comment explains that 'all others' generate both files, but doesn't clarify that init/ contracts ONLY generate the init variant. The comment could be misleading about what files are actually produced for each category.
  optimizer/yul_source_optimizer.py:1
• The regex pattern for matching balanced parentheses is complex and only handles 2 levels of nesting. For deeply nested expressions, this could fail to match. Consider using a recursive approach or documenting the nesting limitation.
  foundry/test/StateTest.t.sol:1
• The path uses ../output/ which assumes the test is run from foundry/ directory. This breaks if tests are run from the project root. Consider using a more robust path resolution approach or documenting the required working directory.
  foundry/test/StateTest.t.sol:1
• The path uses ../output/ which assumes the test is run from foundry/ directory. This breaks if tests are run from the project root. Consider using a more robust path resolution approach or documenting the required working directory.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[GlebRadchenko]

@codex review

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 7bbcd393a1

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Allow nonce 0 when computing sidecar addresses

The guard if not deployer or not main_nonce treats nonce=0 as “missing” and returns early, which skips sidecar address recomputation for the common case of the deployer’s first CREATE. That means compute_sidecar_addresses won’t update auto_predicted for default configs where deployment.nonce is 0, so sidecar addresses stay stale even when Yul order changes. Consider only rejecting None/empty values, not zero.

Useful? React with 👍 / 👎.

[chatgpt-codex-connector]

Runtime-only output can miscompute dataoffsets for multiple sidecars

This appends all sidecar blobs after the runtime and relies on runtime dataoffset() to resolve offsets. But the generator’s runtime path computes dataoffset as codesize - datasize (see VenomIRBuilder._handle_intrinsic), which only yields the start of the last appended blob. With multiple sidecars, any dataoffset("SidecarX") in runtime code will point to the last sidecar instead of the named one. This will break runtime-only tests/contracts that deploy multiple sidecars from runtime using dataoffset/datasize.

Useful? React with 👍 / 👎.