builtin: reclaim captured closure contexts

[GGRei]

Summary

This rescues the closure leak fix from vlang#27446 with a smaller, scoped patch.

Captured closure contexts now use collectable memdup instead of memdup_uncollectable, while the closure runtime keeps live contexts reachable through a scanned table. Local captured closures that do not escape are cleaned up when their scope is left, including normal scope exit, return, and labeled/non-labeled break / continue.

The cleanup is conservative: closures are not destroyed when they escape through return values, copies, function arguments, struct fields, globals, defer blocks, spawn, or go.

Changes

• Add explicit closure context ownership through closure_create_with_data.
• Keep live closure contexts rooted under Boehm while the closure is alive.
• Clear root table entries before deleting them.
• Generate collectable closure contexts for captured anon functions.
• Distinguish owned value receivers from borrowed pointer receivers for method-value closures.
• Add local non-escaping closure cleanup in Cgen.
• Add regression tests for the Capturing closures leak their captured context under -gc boehm (unbounded growth in frame-based apps) vlang/v#27445 memory growth pattern and generated C safety cases.

Tests

./v -o ./v_patch_27445 cmd/v
./v_patch_27445 test vlib/v/gen/c/closure_context_codegen_test.v
./v_patch_27445 -gc boehm test vlib/v/tests/fns/closure_context_boehm_root_test.c.v
./v_patch_27445 -gc none test vlib/v/tests/fns/closure_context_boehm_root_test.c.v
VJOBS=1 ./v_patch_27445 test vlib/v/tests/fns
./v_patch_27445 test vlib/v/gen/c/coutput_test.v
git diff --check

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: e3c1b6c869

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".

[GGRei]

@codex review

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: cc9e5b6381

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".

[GGRei]

@codex review

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 1ca0da1cd4

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".

[GGRei]

@codex review

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: f558787b1a

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".

[GGRei]

@codex review

[chatgpt-codex-connector]

Codex Review: Didn't find any major issues. Chef's kiss.

Reviewed commit: 6c0a119ba2

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".

[GGRei]

@codex review

[MartenH]

@GGRei Have you checked out fat closures as an alternative to this?
RFC_closures.md

[GGRei]

@MartenH Yes, I looked at the fat-closures RFC. I agree it is the cleaner long-term direction: make pure V closures GC-visible and keep trampolines only for C ABI coercions.

I do not think it is a drop-in alternative for this PR, though. This PR intentionally fixes the leak under the existing closure representation. Fat closures would require changing fn value representation/calling convention, Cgen call lowering, C FFI coercions, -gc none/-autofree ownership, bootstrap, and eventually native backend support.

Given where the V1/C backend is in its lifecycle, I would prefer not to block this scoped fix on a much larger closure ABI/codegen redesign. My preference would be land the current fix as an interim unblocker, and treat fat closures as a separate staged RFC/implementation, probably aligned with future V2 ownership/autofree work.

If/when fat closures land, the manual lifetime/reclaim API can be removed or deprecated.