Fix mypy type narrowing false positive in cfg.py

[Copilot]

mypy 1.20.0 CI lint job fails due to type narrowing behavior: when evaluating te.entry is self._current_try_begin, mypy narrows te.entry from TryBegin to TryBegin | None because the RHS (self._current_try_begin) is Optional[TryBegin].

Change

• src/bytecode/cfg.py: Add explicit self._current_try_begin is not None guard before the identity comparison, giving mypy enough information to narrow both sides correctly:

# Before
) and te.entry is self._current_try_begin:

# After
) and self._current_try_begin is not None and te.entry is self._current_try_begin:

Semantically equivalent — the identity check can never be True when self._current_try_begin is None.

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 95.16%. Comparing base (1757c49) to head (29a1e71).

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #190   +/-   ##
=======================================
  Coverage   95.16%   95.16%           
=======================================
  Files           7        7           
  Lines        2028     2028           
  Branches      449      449           
=======================================
  Hits         1930     1930           
  Misses         54       54           
  Partials       44       44           

☔ View full report in Codecov by Sentry.
📢 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.

[MatthieuDartiailh]

TryEnd.entry cannot be None so the error from Mypy is weird. Could you try flipping the test around self._current_try_begin is te.entry to see if it satisfies Mypy ?

[Copilot]

Tried flipping to self._current_try_begin is te.entry — it produces the same 3 errors. The root cause is mypy's identity-narrowing behavior: when evaluating x is y where y is Optional[T], mypy narrows x into T | None in the truthy branch (reasoning: if y were None and x is y is True, then x must also be None). This applies regardless of which side the optional is on.

You're right that TryEnd.entry can never actually be None — but mypy doesn't exploit that to suppress the narrowing. The explicit is not None guard is the minimal fix that teaches mypy self._current_try_begin is non-None before the identity check, preventing the spurious narrowing of te.entry.

[MatthieuDartiailh]

@copilot Did you try of just assume the behavior ? Also be mindful of the formatting which is currently erroring in CI

[Copilot]

I did try it — flipped to self._current_try_begin is te.entry produces the same errors. Mypy's identity narrowing widens the non-optional operand to include None regardless of which side of is it's on, so the is not None guard is required.

The formatting issue has been fixed in commit 7702d3e — black now reformats the walrus-operator if condition and the ternary expression to use the parenthesised multi-line style it expects.

[MatthieuDartiailh]

Address the formatting issue identified by the CI (lint job)

[MatthieuDartiailh]

@copilot fix the formatting issue

[Copilot]

Address the formatting issue identified by the CI (lint job)

Fixed in commit 7702d3e — ran black on cfg.py to apply the required formatting changes.

@copilot fix the formatting issue

Done in commit 7702d3e.