GeezerVison

Copy link

Copy Markdown

Contributor

ncoghlan reviewed Feb 18, 2017

View reviewed changes

berkerpeksag requested changes Feb 19, 2017

View reviewed changes

Carreau force-pushed the undeprecate branch from dbcffb0 to f90fb68 Compare February 20, 2017 23:34

berkerpeksag requested changes Feb 21, 2017

View reviewed changes

Sorry, something went wrong.

Uh oh!

There was an error while loading. Please reload this page.

Carreau force-pushed the undeprecate branch from f90fb68 to e3e6084 Compare February 22, 2017 03:03

berkerpeksag approved these changes Feb 22, 2017

View reviewed changes

Carreau deleted the undeprecate branch February 22, 2017 18:23

…ython#122)

Nick Coghlan said on bpo-28814:

> inspect.getargvalues() and inspect.formatargvalues() were deprecated
> in Python 3.5 as part of implementing bpo-20438

> This is incorrect, as these are *frame* introspection related functions,
> not callable introspection ones. The documentation and implementation
> layout is confusing though, as they're interleaved with the callable
> introspection operation

This commit undeprecates these functions and adds a note to ignore
previous deprecation notices.

(cherry picked from commit 0899b98)

…ython#122)

Nick Coghlan said on bpo-28814:

> inspect.getargvalues() and inspect.formatargvalues() were deprecated
> in Python 3.5 as part of implementing bpo-20438

> This is incorrect, as these are *frame* introspection related functions,
> not callable introspection ones. The documentation and implementation
> layout is confusing though, as they're interleaved with the callable
> introspection operation

This commit undeprecates these functions and adds a note to ignore
previous deprecation notices.

(cherry picked from commit 0899b98)

) (#243)

Nick Coghlan said on bpo-28814:

> inspect.getargvalues() and inspect.formatargvalues() were deprecated
> in Python 3.5 as part of implementing bpo-20438

> This is incorrect, as these are *frame* introspection related functions,
> not callable introspection ones. The documentation and implementation
> layout is confusing though, as they're interleaved with the callable
> introspection operation

This commit undeprecates these functions and adds a note to ignore
previous deprecation notices.

(cherry picked from commit 0899b98)

) (#244)

Nick Coghlan said on bpo-28814:

> inspect.getargvalues() and inspect.formatargvalues() were deprecated
> in Python 3.5 as part of implementing bpo-20438

> This is incorrect, as these are *frame* introspection related functions,
> not callable introspection ones. The documentation and implementation
> layout is confusing though, as they're interleaved with the callable
> introspection operation

This commit undeprecates these functions and adds a note to ignore
previous deprecation notices.

(cherry picked from commit 0899b98)

The cache caused false block leak errors from test suite --huntrleaks / -R runs. The performance penalty is irrelevant in debug builds.

https://bitbucket.org/stackless-dev/stackless/issues/122

Reinstates block_map_n294_chain at 150 arms (n≈302 BBs) — the wiring
fixture that exposed the Phoenix-introduced D-1775660617 RPO un-clamp
bug at push 42 (then HALT'd via SIGABRT). Push 43 (4feb0d2) closed
the bug; this push restores the canary now that it legitimately PASSes.

Test entries: n=0/74/149/200 (low/mid/high-arm + default-miss path
coverage), all asserted force_compile + correct execution.

Coverage role:
  - 110-arm fixture (push 41, n≈223 BBs, no resize) — covers below-256
    code path
  - 150-arm fixture (this commit, n≈302 BBs, resize-triggered) — covers
    >256 code path that GitRPOTraversal + un-clamped wrapper now handle
  - Together: empirical pair on the same fn-shape exercising both paths
    of hir_cfg_get_rpo + the cap=4096 paths in pass_output_type_c.c

Coverage-Trigger Rule (Phoenix-presumed-regression policy, Alex
13:47:22Z + 13:48:14Z; advisory per supervisor 13:18:54Z): re-grep of
parked-bug list (docs/parked-bug-audit-2026-04-24.md Groups A+C; 8 items)
against the n≈302 BB fixture surface returns ZERO newly-reachable items
(see chat 14:42:05Z for per-item disposition). Push 44 is the rule's
first practical test; honoured-in-the-letter at birth.

A/B coverage (pythia python#122 python#3 follow-up) discharged STRUCTURALLY per
theologian 14:41:27Z: GetRPOTraversal is deterministic on CFG state;
the cap parameter controls only caller-buffer write count, not algorithm
output. Same-function pre-sized-vs-resize empirical A/B (option i) would
require a debug-only cap-setter knob (multi-hour infra) and would add
empirical-on-empirical, not new evidence about the determinism claim.
Tie-breaking happens INSIDE GetRPOTraversal which receives the CFG, not
the caller's output buffer; caller-buffer realloc cannot reach that
surface.

Pre-commit discipline: bash -n PASS; no C/C++ source touched (push 43
binary 4feb0d2 remains current per testkeeper 14:42:44Z).

Auth chain: theologian 13:46:04Z (A/B fold-in proposal) + 14:41:27Z
((ii) APPROVE + structural defense); supervisor 14:39:27Z (push 43 +
push 44 plan) + 14:41:00Z (proceed-(ii)-unless-objected); testkeeper
pre-commit lint 14:42:44Z; alex 13:47:22Z + 13:48:14Z policy directives.