Regression report stable-2016-05-24 vs. nightly-2016-06-24

[brson]

Root regressions, sorted by rank:

• typemap-0.3.3 (before) (after)
• bytes-0.3.0 (before) (after)
• cargo-0.10.0 (before) (after)
• simple_parallel-0.3.0 (before) (after)
• rustc-test-0.1.3 (before) (after)
• eventual-0.1.7 (before) (after)
• sprs-0.4.0-alpha.4 (before) (after)
• grabbag_macros-0.1.0 (before) (after)
• nfd-sys-0.1.5 (before) (after)
• floaty-0.1.0 (before) (after)
• gml-1.1.0 (before) (after)
• chamkho-0.0.13 (before) (after)
• aurelius-0.1.8 (before) (after)
• automaton-0.0.1 (before) (after)
• nickel_mustache-0.1.0 (before) (after)
• onionsalt-0.4.2 (before) (after)
• pacthash-0.2.0 (before) (after)
• parry-0.1.0 (before) (after)
• polly-0.1.2 (before) (after)
• remember-1.0.0 (before) (after)
• tokei-3.0.2 (before) (after)
• toml_document-0.1.3 (before) (after)
• accurate-0.2.1 (before) (after)

Full Report.

[eefriedman]

It looks like multiple crates are hitting https://github.com/rust-lang/rust/issues/34431 ; specifically, it looks like the clap crate depends on this construct.

It looks like multiple crates are hitting an error caused by https://github.com/rust-lang/rust/pull/33982 … at least, I’m guessing that’s what it is (errors like `$t:expr` is followed by `{`).

It might be worth specifically trying to get a nickel v0.7.4 and a crossbeam v0.1.7 published to reduce the number of broken crates.

I have no idea what’s causing the error in nfd-sys (“error: failed to learn about crate-type dylib early on”).

Would it be possible to get crater to separate out errors caused by deny(warnings) from other errors, since we don’t really care that a bunch of crates have unused import warnings?