fix(templates): avoid reusing specialization without rescan info in C… by jakub-labenski · Pull Request #390 · edgcpp/compiler

jakub-labenski · 2026-10-26T10:25:52Z

…TAD []

Discussion #336

compare_expressions checked, under CC_EXACT_EQUIVALENCE, that both nodes either have or lack rescan info. That check only ran when the type comparison was done, and the type comparison is skipped for variable references.

As a result, a specialization such as P<b> (b a dependent static data member) created by a member declaration, with no rescan info, could be reused for the same specialization in a constructor parameter type. Then, building the implicit deduction guide for class template argument deduction failed with the internal error "missing default rescan info".

Move the rescan info check out of the type comparison block so it applies to all equivalent nodes.

Add regression test
regressions/ctad-dependent-member-template-arg-reuse.sft.cpp and a Changes entry.

Summary by CodeRabbit

Bug Fixes
- Fixed compilation failures in certain C++17 class template argument deduction cases involving dependent member-template arguments reused within a class.
- Improved expression comparison consistency, including cases where type comparison is skipped.

coderabbitai · 2026-10-26T10:26:46Z

No actionable comments were generated in the recent review.

Recent review info

Run configuration

Source mode: Auto
Repository: Unavailable
Branch: Unavailable
File: No configuration file from this repository applied
Commit: Unavailable
Layers: defaults
Review profile: CHILL
Plan: Advanced
Run ID: 298a87e5-9823-474a-bb8c-570acbb66fc0

Commits

Reviewing files that changed from the base of the PR and between e560a91 and 4d11f84.

Files selected for processing: 7

src/il.c
src/pending_changes/ctad-missing-rescan-info.txt
tests/tests/regressions/.ctad-dependent-member-template-arg-reuse.rto/default.1.1.txt
tests/tests/regressions/.ctad-dependent-member-template-arg-reuse.rto/default.2.1.txt
tests/tests/regressions/.ctad-dependent-member-template-arg-reuse.rto/default.3.1.txt
tests/tests/regressions/.ctad-dependent-member-template-arg-reuse.rto/default.4.1.txt
tests/tests/regressions/ctad-dependent-member-template-arg-reuse.sft.cpp

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

src/pending_changes/ctad-missing-rescan-info.txt

Included review availability: This review used your included allowance. Your plan provides up to 10 included reviews per hour; 9 remain after this review.

Walkthrough

The exact-equivalence check for mismatched rescan information now also runs when type comparison is skipped. A C++17 regression test covers CTAD with reused dependent member template arguments.

Changes: CTAD rescan-info handling

Layer / File(s)	Summary
Equivalence check and CTAD regression `src/il.c`, `tests/tests/regressions/ctad-dependent-member-template-arg-reuse.sft.cpp`, `tests/tests/regressions/.ctad-dependent-member-template-arg-reuse.rto/*`, `src/pending_changes/ctad-missing-rescan-info.txt`	`compare_expressions` now checks mismatched `rescan_info` presence even when type comparison is skipped. The C++17 regression test exercises CTAD for two class templates. Compiler-specific test invocations and the change note describe the case.

Priority: Low

Estimated code review effort: 2 (Simple) | ~10 minutes

Change: Bug fix

Possibly related issues

edgcpp/compiler#395 — Reports the same missing default rescan info error during implicit deduction-guide construction for CTAD.

Suggested reviewers: nv-cmeerw

Merge Risk: Minimal · up to 4d11f

The change addresses the reported CTAD failure; no merge-blocking issue was identified.

Pre-merge checks: 4 · 1

Failed checks: 1 inconclusive

Check name	Status	Explanation	Resolution
Docstring Coverage		Docstring coverage is 0.00% which is insufficient. The required threshold is 80.00%. Docstring coverage is scoped to functions touched by this diff. Documentation-only edits and deletions within survi…	Write docstrings for the functions missing them to satisfy the coverage threshold.

Passed checks: 4 passed

Check name	Status	Explanation
Description Check		Check skipped - CodeRabbit’s high-level summary is enabled.
Title check		The title clearly and concisely describes the main change: preventing reuse of specializations without rescan information during CTAD.
Linked Issues check		Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check		Check skipped because no linked issues were found for this pull request.

Full details: Docstring Coverage

Explanation

Docstring coverage is 0.00% which is insufficient. The required threshold is 80.00%. Docstring coverage is scoped to functions touched by this diff. Documentation-only edits and deletions within surviving functions are included; unchanged diff context and fully deleted functions are excluded. Analyzed 2 functions across 1 files. (6 skipped: 5 unsupported, 1 too large.)

Resolution

Write docstrings for the functions missing them to satisfy the coverage threshold.

Finishing Touches

🧪 Generate unit tests (beta)

Create a new PR

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

_{Comment @coderabbitai help to get the list of available commands.}

edgcpp-bot · 2026-10-26T12:26:12Z

AcknowlEDG review results updated for e560a91f4de1:

Test - Changes (new)

edgcpp-bot · 2026-10-26T12:40:05Z

AcknowlEDG review results updated for e560a91f4de1:

Bench - C (new)
Bench - C++ (new)
Test - Changes (unchanged)

CI success.

codecov · 2026-10-26T12:41:08Z

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

nv-cmeerw

Thanks for working on this, just a few nits.

nv-cmeerw · 2026-10-27T09:01:35Z

@@ -0,0 +1,17 @@
+[GH #335]


That's GH #395 now.

nv-cmeerw · 2026-10-27T09:08:19Z

@@ -0,0 +1,17 @@
+[GH #335]
+Fix CTAD rescan info for reused specializations


Usually, the summary line would say something like "Abort ..." for these kind of issues, and then also mention the function (get_expr_rescan_info) in the description. EDGcpfe/26631, EDGcpfe/23606,EDGcpfe/23607, and EDGcpfe/22589,EDGcpfe/23473 described similar issues.

nv-cmeerw · 2026-10-27T09:08:47Z

    }  /* if */
+    if (eq && (options & CC_EXACT_EQUIVALENCE) &&
+        (node1->extra.rescan_info != NULL) !=
+                                       (node2->extra.rescan_info != NULL)) {


minor nit: the continuation line should be right-aligned at column 79.

nv-cmeerw · 2026-10-27T09:12:03Z

@@ -0,0 +1,31 @@
+//type:fp
+//options_all:--c++17


I'd usually also run the test in common emulation modes, e.g.

//options:--c++17:--gnu_version 150200:--clang_version 220100:--ms_c++17 --microsoft_version 1950

…TAD [GH edgcpp#395] compare_expressions checked, under CC_EXACT_EQUIVALENCE, that both nodes either have or lack rescan info. That check only ran when the type comparison was done, and the type comparison is skipped for variable references. As a result, a specialization such as P<b> (b a dependent static data member) created by a member declaration, with no rescan info, could be reused for the same specialization in a constructor parameter type. Then, building the implicit deduction guide for class template argument deduction failed with the internal error "missing default rescan info". Move the rescan info check out of the type comparison block so it applies to all equivalent nodes. Add regression test regressions/ctad-dependent-member-template-arg-reuse.sft.cpp and a Changes entry.

edgcpp-bot · 2026-10-27T12:00:29Z

AcknowlEDG review results updated for 4d11f84cb0f4:

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

CI success.

nv-cmeerw requested changes Oct 27, 2026

View reviewed changes

jakub-labenski force-pushed the fix/ctad-rescan-info-equivalence branch from e560a91 to 4d11f84 Compare October 27, 2026 10:09

nv-cmeerw approved these changes Oct 27, 2026

View reviewed changes

Navigation Menu

Repository navigation

fix(templates): avoid reusing specialization without rescan info in C… - #390

fix(templates): avoid reusing specialization without rescan info in C…#390
jakub-labenski wants to merge 1 commit into
edgcpp:mainfrom
jakub-labenski:fix/ctad-rescan-info-equivalence

jakub-labenski commented Oct 26, 2026 •

edited by coderabbitai Bot

Loading

Uh oh!

coderabbitai Bot commented Oct 26, 2026 •

edited

Loading

Uh oh!

edgcpp-bot commented Oct 26, 2026

Uh oh!

edgcpp-bot commented Oct 26, 2026

Uh oh!

codecov Bot commented Oct 26, 2026

Uh oh!

nv-cmeerw left a comment

Uh oh!

nv-cmeerw Oct 27, 2026

Uh oh!

nv-cmeerw Oct 27, 2026

Uh oh!

nv-cmeerw Oct 27, 2026

Uh oh!

nv-cmeerw Oct 27, 2026

Uh oh!

edgcpp-bot commented Oct 27, 2026

Uh oh!

This branch has not been deployed

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

		@@ -0,0 +1,17 @@
		[GH #335]
		Fix CTAD rescan info for reused specializations

		@@ -0,0 +1,31 @@
		//type:fp
		//options_all:--c++17

Conversation

jakub-labenski commented Oct 26, 2026 • edited by coderabbitai Bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary by CodeRabbit

Uh oh!

coderabbitai Bot commented Oct 26, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

edgcpp-bot commented Oct 26, 2026

Uh oh!

edgcpp-bot commented Oct 26, 2026

Uh oh!

codecov Bot commented Oct 26, 2026

Codecov Report

Uh oh!

nv-cmeerw left a comment

Choose a reason for hiding this comment

Uh oh!

nv-cmeerw Oct 27, 2026

Choose a reason for hiding this comment

Uh oh!

nv-cmeerw Oct 27, 2026

Choose a reason for hiding this comment

Uh oh!

nv-cmeerw Oct 27, 2026

Choose a reason for hiding this comment

Uh oh!

nv-cmeerw Oct 27, 2026

Choose a reason for hiding this comment

Uh oh!

edgcpp-bot commented Oct 27, 2026

Uh oh!

This branch has not been deployed

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

jakub-labenski commented Oct 26, 2026 •

edited by coderabbitai Bot

Loading

coderabbitai Bot commented Oct 26, 2026 •

edited

Loading