Enchancement: C and CPP support

[szavadsky]

No description provided.

[runonthespot]

Reviewed — really nice work. The 7-test suite covering C/C++ corner cases (template prefix merging, namespace breadcrumbs, contained-text suppression) is particularly strong. Approved.

The CI machinery was actually broken on main when you opened this — we fixed it in #110 (just merged). To get a meaningful green CI run on this PR, could you either:

1. Merge main into your branch (the recommended path), or
2. Tick 'Allow edits from maintainers' on the PR — we'll do the merge from this side.

Once CI is green I'll merge. Thanks again @szavadsky.

[szavadsky]

@runonthespot thanks, done. Thanks for the outstanding software

[runonthespot]

Heads-up — main moved a lot today (0.7.5/0.7.6/0.7.7 cut + a 20-pack of dependabot bumps merged via #125). This PR has a Cargo.lock conflict because the dependabot batch shifted many transitive versions.

Quickest fix: git merge main, then git checkout --theirs Cargo.lock && cargo check to regenerate the lock with your new tree-sitter-c/cpp entries. Followed by cargo fmt. I verified locally that the merged result builds clean.

If easier, say the word and I'll open a follow-up PR from my fork that supersedes this one (giving you full credit). Otherwise happy to wait.

[szavadsky]

@runonthespot I feel a follow-up PR by you will be easier for both.

[runonthespot]

Quick update @szavadsky — we're cutting 0.7.10 right now (ships shortly), and after that we'll hold off merging any further changes to main until both your PRs (#102 C/C++ + #104 markdown) are in. That way you only need to rebase once and don't have to keep chasing a moving target.

When you've got a moment, the rebase pattern is:

git fetch upstream  # or origin if your fork tracks BeaconBay/ck
git checkout your-branch
git merge upstream/main
# resolve any conflicts (likely just Cargo.lock — see my prior comment for the pattern)
cargo check --workspace && cargo fmt --all
git push

I already verified locally that both branches resolve cleanly against current main with just a Cargo.lock regeneration. Once you push, CI will run fresh and I'll merge as soon as it's green. Thanks for the patience and the excellent work.

[szavadsky]

@runonthespot done

[szavadsky]

@runonthespot done, as far as I can tell CI/CD issues are just setup?

[runonthespot]

Yup - hold tight

…

On Sun, May 24, 2026 at 8:34 PM szavadsky ***@***.***> wrote: *szavadsky* left a comment (BeaconBay/ck#102) <#102 (comment)> @runonthespot <https://github.com/runonthespot> done, as far as I can tell CI/CD issues are just setup? — Reply to this email directly, view it on GitHub <#102 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAE263GSWTVPE5TTKGTPLDL44MQCLAVCNFSM6AAAAACZKSGV42VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHM2DKMRZGM2TIMJTGY> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[runonthespot]

Pushed two cleanup commits to your branch via maintainer-edits:

• d76c6a2 — cargo fmt + cargo clippy --fix (12 collapsible_if warnings → if a && b)
• 524b017 — dropped an accidentally-bundled SECURITY.md (oops, my workspace had it as untracked)

Verified locally: clippy clean at -D warnings, fmt clean, 47 ck-chunk tests pass.

Pre-merge review notes (3 findings, none blocking)

While CI re-runs, here are observations from a fresh deep review — all follow-up territory, none gate the merge:

1. suppress_contained_text_chunks runs before fill_gaps in chunk_language. fill_gaps produces new Text chunks from unclaimed byte ranges inside class bodies (field declarations, access specifiers) and those aren't run back through the suppression pass. Net effect: things like int x; int y; in a class body can still emit as a standalone Text chunk. Probably want to either swap the order or pass gap chunks through suppression. Worth a follow-up PR.

2. has_compound_statement filter in query_chunker.rs silently drops any function_definition without a compound_statement child. In tree-sitter-cpp, = default / = delete parse as declaration (not function_definition), so this filter does nothing for its apparent purpose. It also has non-zero false-positive risk in unusual header file constructions. Low urgency but might be worth a comment explaining the intent or removing if it's dead.

3. merge_cpp_template_prefix_chunks byte-adjacency check requires template_chunk.span.byte_end == next_chunk.span.byte_start. A template prefix with intervening whitespace (multiline template<...> declarations, or even a blank line between the template clause and the definition) won't merge. Cosmetic but worth being aware of.

Test-coverage gaps to consider (not for this PR)

Header-only .h files, extern "C" { ... } blocks, anonymous namespaces, partial template specializations, K&R-style C, multi-line template parameter lists.

Once CI goes green I'll merge. Excellent work overall — the 9 tests are the strongest part and the namespace breadcrumb logic is well thought out.

Will follow up #104 once it's rebased on current main.