[yangfan-yf-yf]

I found that positional-only and keyword-only parameters can resolve to an outer assignment instead of their own binding. Inlining target = 42 above def func(*, target): return target generates the invalid declaration def func(*, 42). Inlining only the body reference leaves valid code that returns 42 instead of the int passed by func(target=int). Rename and argument inference also lose the parameter identity or use the wrong argument slot.

I now give each parameter a stable binding slot, distinguish positional and keyword matching, and evaluate defaults in the enclosing scope. The ordinary slots are positional-only, positional-or-keyword, and keyword-only parameters; variadic object slots follow them. A keyword with the same name as a positional-only parameter goes to **kwargs without replacing the positional slot. Unknown unpacking and unknown supplied values do not invent default bindings.

I also align static analysis, dynamic call records, method receivers, and callable-object receivers with that layout. A receiver is inserted once: static methods receive none, class methods retain the actual dispatch class, and variadic receivers do not occupy keyword-only slots. Versioned, persistable call scopes isolate changed layouts and receiver rules from old records, including fallback reads; ordinary plain-function and per-name records retain their existing scopes. Static-method parameter attributes are kept separate from the containing class, including actual builtin decorator aliases.

Read-only completion and calltips use the parameter metadata while preserving ordinary signature display. Signature-writing refactorings reject unsupported positional-only or keyword-only rewrites before generating changes. Existing positional-only Inline Method behavior is preserved; keyword-only Inline Method remains unsupported. This change covers parameter bindings and defaults; general annotation-expression scope and deferred-evaluation inlining remain separate boundaries.

Validation:

• 113 new regression cases exercise public refactoring APIs, identity, real program execution, argument inference, dynamic records, persistence, and compatible or rejected signature consumers.
• Core 64 cases pass on Python 3.12 and 3.14; the latest 17-case increment has 15 failures and 2 passing controls on unchanged master. Earlier baseline increments are retained with their original test hashes.
• Python 3.12 full suite: 2262 passed, 12 skipped, 5 xfailed. Local product patch coverage is 334/343 executable added lines, 97.38%; the remaining uncovered lines are recorded separately.
• Python 3.14 full suite: 2266 passed, 8 skipped, 5 xfailed.
• All five repository style hooks pass on the 21 changed files. Both new test files also pass explicit formatting/import checks, and git diff --check passes. Product, test, and documentation hashes are unchanged across the final runs and checks.

[codecov]

Codecov Report

❌ Patch coverage is 99.06152% with 9 lines in your changes missing coverage. Please review.
✅ Project coverage is 95.47%. Comparing base (22f5e50) to head (7733d55).

Files with missing lines	Patch %	Lines
rope/contrib/codeassist.py	95.52%	3 Missing ⚠️
rope/base/pyobjects.py	50.00%	2 Missing ⚠️
rope/base/oi/objectinfo.py	98.27%	1 Missing ⚠️
rope/refactor/functionutils.py	92.30%	1 Missing ⚠️
rope/refactor/inline.py	94.73%	1 Missing ⚠️
ropetest/parameter_kinds_test.py	99.79%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #898      +/-   ##
==========================================
+ Coverage   95.34%   95.47%   +0.12%     
==========================================
  Files         134      136       +2     
  Lines       26762    27667     +905     
==========================================
+ Hits        25516    26414     +898     
- Misses       1246     1253       +7     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[marlon-costa-dc]

Independently arrived at the same parameter-binding fixes from a different repro; after diffing against this PR we found nothing functional on our side that #898 does not already cover, so we are adopting #898 as-is.

We did port three edge-case parametrizations our repro had that the current suite here does not: a signature with positional-only plus keyword-only but no *args, a method whose only parameter is keyword-only (receiver must not be injected), and saved object analysis for a call mixing positional, *args, and keyword-only arguments. They live at dc/pr-898 (one commit on top of this branch, parameter_kinds_test.py 67 passed on 3.13) — feel free to cherry-pick them into this PR if you agree they add coverage.