[nv-cmeerw]

Fixes #500

A pack expansion in a generic lambda nested in a template instantiation could
produce an error node in the IL when it referenced a lambda parameter pack
before an empty enclosing parameter pack. Empty enclosing packs are now
handled correctly regardless of the order of the pack references. Spurious
warnings from prototype instantiations in such cases are also suppressed. For
example, with --c++20 -tused:

  template<int... Is> void f() {
    []<int... Js>() { ((Js * Is) * ... * 1); };
  }
  template void f<>();  // Previously produced an error node in the IL.

Summary by CodeRabbit

• Bug Fixes
  • Corrected C++20 pack expansion handling in generic lambdas when an enclosing parameter pack is empty, including when pack references appear in different orders.
  • Suppressed spurious warnings during prototype instantiations while preserving relevant warnings during real instantiations.

[coderabbitai]

No actionable comments were generated in the recent review.

Walkthrough

The compiler now sets generic-lambda pack expansion length to zero when the current pack is empty in a real instantiation. It also suppresses a warning in the related instantiation case. Regression tests cover pack combinations, warning output, and updated diagnostics.

Changes: Empty pack handling

Layer / File(s)	Summary
Expansion length and warning handling src/scope_stk.c, src/scope_stk.h, src/expr.c	The compiler sets expansion length to zero when the current pack is empty in a generic-lambda real instantiation. It exposes the instantiation check and uses it to suppress a warning in the specified case.
Regression coverage tests/tests/regressions/gh-500*.sft.cpp, tests/tests/regressions/.gh-500*.rto/*, tests/tests/imported/clang/cpp/SemaCXX/.lambda-pack-expansion.rto/*, tests/tests/imported/gnu/cpp/cpp1z/*.rto/*, src/pending_changes/gh-500.txt	Regression tests cover empty and nonempty pack combinations and warning output. Imported Clang expected output removes two errors in each updated result. Imported GNU expected output updates instantiation traces. The pending-changes note describes the GH #500 case.

Estimated code review effort: 3 (Moderate) | ~20 minutes

Change: Bug fix · Severity of issue fixed: Low

Possibly related PRs

• edgcpp/compiler#492: Suppresses missing-parameter-pack diagnostics for nested generic-lambda pack expansions and updates expected output for overlapping pack-expansion cases.

Suggested labels: bug

Suggested reviewers: daveedvdv-nvidia

Merge Risk: Minimal · up to c5dc4

The change fixes an error node in the IL for generic lambda pack expansions that reference an empty enclosing pack, and it adds regression tests. No concrete merge-blocking risk was identified.

Pre-merge checks: 4

Check name	Status	Explanation
Description Check		Check skipped - CodeRabbit’s high-level summary is enabled.
Title check		The title "Empty enclosing packs in generic lambdas [GH #500]" matches the main change. The raw summary and objectives describe a fix for an error node in the IL when a generic lambda's pack expansion…
Linked Issues check		Requirement [#500]: a pack expansion in a generic lambda nested in a real instantiation must not produce an enk_error node or tk_error type when an empty enclosing pack is referenced before a lambda p…
Out of Scope Changes check		Changes map to #500: the src/scope_stk.c fix (empty-pack handling in hybrid pack expansion), the linkage change that exposes in_generic_lambda_in_real_instantiation() and its src/scope_stk.h declarati…

Full details: Title check

Explanation

The title "Empty enclosing packs in generic lambdas [GH #500]" matches the main change. The raw summary and objectives describe a fix for an error node in the IL when a generic lambda's pack expansion references an empty enclosing parameter pack, and the title names that same subject. It is specific and concise, and it names the linked issue. No search was needed for this check.

Full details: Linked Issues check

Explanation

Requirement [#500]: a pack expansion in a generic lambda nested in a real instantiation must not produce an enk_error node or tk_error type when an empty enclosing pack is referenced before a lambda pack. The whole-PR diff of src/scope_stk.c (merge base to head) adds, in the hybrid-expansion branch, a rule that sets elements = 0 when elements_for_pack == 0. This applies regardless of which pack is encountered first, which matches the reproducer ((Seq * Vals) * ... * 1 with Vals = <>). The whole-PR diff of src/expr.c adds a suppression of the 'expression has no effect' warning when in_generic_lambda_in_real_instantiation() is true. Regression tests under tests/tests/regressions/ (gh-500.sft.cpp, gh-500.rto, gh-500-warnings.sft.cpp) cover the empty-pack case. The incremental diff from the previous reviewed head (cfe283b) to the current head (c5dc46e) could not be read, so changes made in that increment are not verified. The whole-PR diffs of the two source files match the previous assessment. Limits: I did not read the .rto expected-output files directly; the result relies on the summary for them.

Full details: Out of Scope Changes check

Explanation

Changes map to #500: the src/scope_stk.c fix (empty-pack handling in hybrid pack expansion), the linkage change that exposes in_generic_lambda_in_real_instantiation() and its src/scope_stk.h declaration (needed by src/expr.c), the src/expr.c warning suppression for generic lambdas in real instantiations (same code path as the reproducer), the pending-changes note for #500, and the gh-500 regression tests. The imported clang .lambda-pack-expansion.rto expectations drop two GH112352::S::foldExpr2 template-template-parameter errors. These are expected-output side effects of the same pack-expansion code path, not a demonstrated unrelated change. A reviewer should still confirm that removing these diagnostics is correct and does not hide a real error. The incremental diff since the previous reviewed head could not be read, so its content is not verified. No change is shown to be unrelated to #500.

Finishing Touches

Generate unit tests (beta)

• Commit to this branch
• Create a new PR

• Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

Share

• X
• Mastodon
• Reddit
• LinkedIn

Recent review info

Run configuration

• Source mode: Auto
• Repository: edgcpp/compiler
• Branch: main
• File: .coderabbit.yaml
• Commit: 63d8840
• Layers: Repository YAML
• Review profile: CHILL
• Plan: Advanced
• Run ID: d748db38-bf22-41c5-bd7e-7c49b30b2c4d

Commits

Reviewing files that changed from the base of the PR and between cfe283b and c5dc46e.

Files selected for processing: 19

• src/expr.c
• src/pending_changes/gh-500.txt
• src/scope_stk.c
• src/scope_stk.h
• tests/tests/imported/clang/cpp/SemaCXX/.lambda-pack-expansion.rto/default.1.1.txt
• tests/tests/imported/clang/cpp/SemaCXX/.lambda-pack-expansion.rto/edg_x86_32_cp.1.1.txt
• tests/tests/imported/clang/cpp/SemaCXX/.lambda-pack-expansion.rto/edg_x86_64_cp.1.1.txt
• tests/tests/imported/gnu/cpp/cpp1z/.constexpr-if-lambda5.rto/default.1.1.txt
• tests/tests/imported/gnu/cpp/cpp1z/.fold-lambda4.rto/default.1.1.txt
• tests/tests/regressions/.gh-500-warnings.rto/default.1.1.txt
• tests/tests/regressions/.gh-500-warnings.rto/default.2.1.txt
• tests/tests/regressions/.gh-500-warnings.rto/default.3.1.txt
• tests/tests/regressions/.gh-500-warnings.rto/default.4.1.txt
• tests/tests/regressions/.gh-500.rto/default.1.1.txt
• tests/tests/regressions/.gh-500.rto/default.2.1.txt
• tests/tests/regressions/.gh-500.rto/default.3.1.txt
• tests/tests/regressions/.gh-500.rto/default.4.1.txt
• tests/tests/regressions/gh-500-warnings.sft.cpp
• tests/tests/regressions/gh-500.sft.cpp

Files skipped from review as they are similar to previous changes: 1

• src/pending_changes/Spurious error nodes in IL for empty pack expansions #500.txt

Included review availability: This review used your included allowance. Your plan provides up to 4 included reviews per hour; 3 remain after this review.

[edgcpp-bot]

AcknowlEDG review results updated for cfe283b811fe:

• Test - Changes (new)

[edgcpp-bot]

AcknowlEDG review results updated for c5dc46e037e9:

• Bench - C (new)
• Bench - C++ (new)
• Test - Changes (updated)

CI success.

[daveedvdv-nvidia]

LGTM.