Interpreter bypasses overloaded logical negation

オープン
#2,122 コメント 1 件 リアクション 0 件 担当者 0 名 GitHub で見る

メンテナーはふだん 1 日以内に返信

まだ誰も着手していません。

評価

難易度
4/5
見積もり時間
3〜5日
初心者へのやさしさ
47/100
issue の種類
バグ
明瞭さ
明確に書かれている
活発さ
活発
技術スタック
java, perl
領域
backend, testing

調査の方向性

Start with CompileOperator.visitOperator, the interpreter’s NOT handler, and MathOperators.not()/Overload.bool_not() to trace how overloaded negation differs across backends. First run the focused reproducer and the unchanged Math-Logic 1.19 t/logic.t suite on system Perl, the JVM backend, and the interpreter. Done means a project regression passes on both backends, raw negation still works under no overloading, and all 332 upstream assertions pass.

索引モデルが issue の本文から書いたものです。

説明

area:backend area:cpan-port area:runtime bug

Summary

Math::Logic 1.19's upstream t/logic.t fails five assertions under PerlOnJava when negating overloaded objects with !. System Perl passes the complete suite. A focused reproducer confirms that the PerlOnJava interpreter returns a plain boolean instead of invoking the class's overloaded ! method.

Evidence

  • CPAN random-tester run: 20261119-133823-51077.
  • Distribution: pure-Perl Math-Logic 1.19; no XS/native dependency.
  • System Perl 5.42.2: t/logic.t passes all 332 tests.
  • PerlOnJava: 5/332 tests fail (tests 230, 260, 278–280), all in ! assertions for three-valued or multi-valued logic. The same failures reproduce with the JVM-default launcher and --interpreter on source 543f9dfef.
  • Focused reducer: system Perl and the JVM backend return a Math::Logic object for the overloaded negation; the interpreter returns a scalar and fails the result-type/value assertion.
use Math::Logic qw($UNDEF);
my $x = Math::Logic->new(-value => $UNDEF, -degree => 3, -propagate => 1);
my $negated = !$x;
die "expected Math::Logic result\n"
    unless ref($negated) eq 'Math::Logic' && $negated == $UNDEF;

Root cause

The interpreter compiler emits Opcodes.NOT when overloading is enabled (CompileOperator.visitOperator). The interpreter's NOT handler currently calls RuntimeScalar.getBoolean() and stores a cached plain true/false scalar. It does not route through MathOperators.not() or Overload.bool_not(), so it skips the (! overload. NOT_NO_OVERLOAD is a separate opcode and should remain the raw-boolean path.

The JVM implementation routes logical negation through MathOperators.not(), which calls Overload.bool_not() for overloaded values. The focused JVM reducer passes, but the complete upstream test still fails five assertions through the default launcher. That state-dependent full-suite failure must also be reduced and resolved before claiming distribution compatibility.

Acceptance criteria

  • Add a project-owned regression for overloaded !, validate it with system Perl first, and make it pass on the JVM and interpreter.
  • Preserve raw boolean behavior under no overloading.
  • Reduce and resolve the five full-suite failures, including the discrepancy between the passing focused JVM case and failing t/logic.t run.
  • Pass the unchanged Math-Logic 1.19 t/logic.t suite (332/332) on both PerlOnJava backends.

Related issue search

No existing issue was found for overloaded unary !. #2074 covers separate Set::Scalar parser and binary-operation failures; #2007 covers recursive re-dispatch of an active overload handler.

主要言語
Perl
スター
67
フォーク
8
平均マージ
5時間 12分
マージ済み PR(30日)
131

環境構築

はじめの一歩

  1. issue を最後まで読み、次にプロジェクトのコントリビューションガイドを読みます。
  2. 着手することを issue にコメントします — 二人が同じ作業をするのを防げます。
  3. リポジトリをフォークし、ブランチを切って変更します。
  4. issue 番号を参照したプルリクエストを送ります。

fglock/PerlOnJava のほかの issue

fglock/PerlOnJava の issue をすべて見る

似ている issue

Perl の issue をもっと見る

新しい issue をメールで受け取る

初心者向けの GitHub issue を短くまとめたダイジェスト。