Ruby

Fix condition, 0 is truethy in Ruby [ci skip]

Bump taiki-e/install-action

Bumps the github-actions group with 1 update in the / directory: taiki-e/install-action.

Updates taiki-e/install-action from 2.71.1 to 2.71.2

• Release notes...

Fix bug reference number and tracker ID to match Bug #21342

Fix typo "rescured" to "rescued"

Revert "ZJIT: Skip TestOpenURI and TestOpenURISSL (#16603)"

This reverts commit 33b9d2ab9f45a0c66665fb8dc40ddebd07c7b7b7.

Revert "ZJIT: Skip TestOpenURIProxy on Linux (#16606)"

This reverts commit d368d42ec6b50cece947556a2185ce73fa493bfe.

YJIT: ZJIT: Use opt-level=1 on dev builds to avoid SystemStackError

Despite adding rb_ec_stack_check() to Rust code entry points, we've seen
SystemStackError causing test failures due to debug builds consume too
much native stack space. Let's use opt-level=1 so rustc/LLVM are more...

ZJIT: Avoid transmute dodging rustc/LLVM int-to-ptr miscompilation

The transmute triggered a miscompilation on Rust 1.85.0, where the
callback was "optimized" to unconditional go to a SIGTRAPping
instruction.

While later Rust versions don't have the same miscompilation, it seems...

Fix random timeout in test_thread_join_during_finalizers on Windows (#16650)

Reduce the number of spawned processes from 50 to 20 and increase the
timeout from 30s to 60s. Process creation is expensive on Windows, and
the combination caused the test to exceed the 30s limit on Windows CI.

Relax RSS limit in test_named_structs_are_not_rooted

The default limit of 2.0 is too tight for RSS measurements on macOS,
causing random CI failures when the ratio lands just above 2.0 (e.g.
2.05). Increase to 2.2 to account for normal RSS measurement noise....