[Mrun25]

A nonempty iterable passed to pdeque(..., maxlen=0) currently remains in the deque. The constructor's t[-maxlen:] becomes t[0:], retaining every value. Subsequent appends can then grow the supposedly zero-capacity deque.

Use an empty prefix slice for zero bounds while retaining the existing suffix slice for positive bounds. The iterable is still consumed, False retains its accepted integer-zero behavior, and non-integral zero bounds still raise TypeError.

Add six parameterized regressions covering empty, singleton and multi-element generator inputs with 0 and False, input consumption, length, retained maxlen, both append/extend directions and pickle round trips. Also pin rejection of maxlen=0.0. Four new cases fail on the unchanged source; the two empty-input controls pass.

Validation on Linux:

• CPython 3.10.12 and 3.13.16, compiled C extension: 643 passed, 1 existing skip each.
• Both versions with PYRSISTENT_NO_C_EXTENSION=1: 542 passed, 102 existing C-extension skips each.
• Module doctests: all 80 passed on both versions.
• Wheel and sdist build on CPython 3.10; git diff --check: passed.

The existing Hypothesis collection warning remains. Other Python versions, PyPy, platforms, hosted CI and the memory-profile job were not run locally.

Related to #318, which was closed without a fix. This does not change the remove/reverse maxlen behavior or the separate slicing/rotation proposals.

AI-assisted investigation, implementation and testing.

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has insufficient remaining quota.