Deliver defaulted positional-only hook arguments positionally

[adenzhou1350]

Fixes #774.

Hookimpl arguments with defaults are forwarded as keywords, including positional-only arguments. An explicit call value therefore raises TypeError in an otherwise valid hook or wrapper.

Remember optional positional-only names/defaults during registration and deliver them positionally. Fill gaps before a supplied positional-only value, but leave trailing omitted defaults implicit. Keep required-argument validation and hookspec/implementation default precedence unchanged; no keyword-only support change.

Regression coverage includes plain hooks, both wrapper styles, bound methods, callable/decorated call_extra implementations, falsy values, mixed signatures and default gaps.

Validation:

• Unmodified main: 219 existing tests pass on Windows/Linux; the expanded 27-case matrix has 16 failures and 11 passing controls on each.
• After the fix: full suite 246 passed on Windows CPython 3.12/3.14 and Linux CPython 3.12.
• Complete repository pre-commit hooks, including mypy, pass on Linux.

Environment note: the committed uv lock selects pytest 3.2.5, which cannot start on Python 3.12. Linux tests used a task-private editable install and pytest 9.1.1 via uv run --no-sync --no-default-groups python -m pytest; Windows used the native interpreter and isolated package targets because local venv executables are blocked. No supported-Python matrix or performance claim.

AI-assisted with OpenAI Codex; the implementation and tests were independently checked against the unmodified source.

Coverage follow-up: add native class-constructor call_extra cases and express the required default prefix without an unreachable loop-exhaustion branch. Both changed production modules now have 100% statement/branch coverage locally; no coverage threshold or exclusion changed. Upstream matrix/package/benchmark checks passed on the initial revision; reruns for the updated head are pending.

[RonnyPfannschmidt]

🤖 Written by Claude Opus 5.5 via Claude Code for the pluggy maintainers; I prompted it, it did the work, I read it.

Approving. I measured the runtime cost against main before approving. The setup was CPython 3.13, a micro-benchmark, 2 rounds, and the best of 15 runs for each case:

Case	main	this PR	Δ
Hook call, 10 impls, no defaulted args	~4.3 µs	~4.35 µs	noise
Hook call, 10 impls with b=None, c=None	~7.2 µs	~7.9 µs	+8–12% (~60–80 ns per impl)
Hook call, 10 impls with positional-only defaults	TypeError	~15.2 µs	new path, ~1.2 µs per impl
HookImpl() construction	~2.55 µs	~4.35 µs	+70% (~1.8 µs per impl)

• Impls without defaulted args are unaffected, because all the new code is inside if hook_impl.kwargnames:.
• Impls with defaulted args now pay on every call for the _posonlydefaults truthiness check. They also pay for a not in _posonlydefaults lookup for each kwarg inside the kwargs comprehension.
• The registration cost is one-time and negligible in absolute terms.

Follow-up (not blocking, will be folded into a refactoring PR stack): move the split to HookImpl.__init__. That means precomputing a kwargnames tuple without the positional-only names for the kwargs comprehension, plus a precomputed (name, default) sequence for the positional-only fill. The call path then only touches the positional-only branch when it is non-empty. With that, impls with ordinary defaults should cost the same as on main, and only positional-only impls pay extra.

---

Generated by Claude Code