Regression report stable-2015-08-06 vs. nightly-2015-08-27

[brson]

Coverage

• 2886 crates tested: 1478 working / 1062 broken / 15 regressed / 175 fixed / 156 unknown.

Regressions

• There are 7 root regressions
• There are 15 regressions

Root regressions, sorted by rank:

• nickel-0.6.0 (before) (after)
• piston-gfx_texture-0.3.0 (before) (after)
• glium_text-0.5.1 (before) (after)
• lux-0.0.2 (before) (after)
• netlib-blas-provider-0.0.8 (before) (after)
• cuckoofilter-0.2.0 (before) (after)
• rustache-0.0.1 (before) (after)

Non-root regressions, sorted by rank:

• piston_window-0.17.0 (before) (after)
• piston2d-gfx_graphics-0.6.0 (before) (after)
• aurelius-0.1.2 (before) (after)
• confusion-0.0.1 (before) (after)
• cryptosphere-0.0.0 (before) (after)
• nickel_sqlite-0.0.1 (before) (after)
• piston3d-gfx_voxel-0.2.0 (before) (after)
• skeletal_animation-0.1.1 (before) (after)

Full Report.

Analysis

Regressions in some key crates.

• cannot infer lifetime, conflicting requirements - nickel
• ICE! - piston_gfx-texture, glium_text, lux
• timeout (false positive?) - netlib-blas-provider
• unresolved std::hash (bugfix?) - cuckoofilter
• unused_mut (lint change) - rustache

Anybody know what PR caused the std::hash::hash resolve change? Presumably std::hash::hash never existed.

Is there an issue for the ICE? It should be high priority.

cc @nikomatsakis

[bluss]

cuckoofilter is importing std::hash::hash but not using it. This unfortunately made removing that function (it was removed) a breaking change.

Proof on playpen

Sneaking in unstable imports is possible if it’s in an import list – use std::hash::{Hash, hash};

Issue #28075

[nikomatsakis]

@brson I can’t reproduce that ICE on the current master.

[brson]

Great! I’ll get another report out asap.