[do not merge] Relicensing rewrite

[Manishearth]

From the relicensing meta-issue (#3093) we have consent to relicense from all copyright holders except for:

• #2814
• #427
• #3178 (this is a new PR and I hope to get consent from them now)
• #423 (this code has since been rewritten and is minor)

There may be people we have missed, before the final relicense we'll do a proper check.

Anyway, #2814 and #427 need to be rewritten for us to be able to relicense. Ideally, they should be written by someone who has not seen their code.

This PR removes the lints introduced in those PRs. Someone who has not seen those PRs (or this one) needs to rewrite them from scratch. I'll attempt to describe them:

• map_clone (style): catches .map(|x| x.clone()) and .map(|x| *x) and .map(|&x| x), suggests .cloned() (ensure there are no adjustments!)
• fn_to_numeric_cast (style, ideally in CastPass): Cast a function to something other than a usize
• fn_to_numeric_cast_with_truncation (correctness): Same as above, but the target type is smaller.

Make these changes as PRs to this branch, and we'll land it all together.

cc @oli-obk @phansch @flip1995

@oli-obk / @flip1995 reviewed the fn_to_numeric_cast* lints so shouldn't be the ones to re-add them.

Anyone up for writing these lints? :)

[oli-obk]

I'll do map_clone. I don't remember ever touching it and I have not looked at the changes of this PR, just at your post.

[Manishearth]

newline at eof

[Manishearth]

ensure that there aren't any adjustments on path

[phansch]

I haven't seen the code of #2814. Did it add both fn_to_numeric_cast and fn_to_numeric_cast_with_truncation?
I will start with fn_to_numeric_cast later today.

[Manishearth]

Yes and yes. They're really the same lint with an additional check for one :)

…

On Tue, Oct 2, 2018, 4:11 PM Philipp Hansch ***@***.***> wrote: I haven't seen the code of #2814 <#2814>. Did it add both fn_to_numeric_cast and fn_to_numeric_cast_with_truncation? I will start with fn_to_numeric_cast later today. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#3251 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABivSJ3aMZo3gCk-I-koYTuQvMLFTeJVks5ug3QBgaJpZM4XDzll> .

[phansch]

Ok, I have a basic version of fn_to_numeric_cast ready. But I'm not sure if I need to handle only FnDef or also FnPtr and the difference between those two. I guess a code example could be helpful.

I currently have the following test cases:

fn foo() -> String { String::new() }

fn test_function_to_numeric_cast() {
    let _ = foo as i8;
    let _ = foo as i16;
    let _ = foo as i32;
    let _ = foo as i64;
    let _ = foo as i128;
    let _ = foo as isize;

    let _ = foo as u8;
    let _ = foo as u16;
    let _ = foo as u32;
    let _ = foo as u64;
    let _ = foo as u128;

    // Casting to usize is OK and should not warn
    let _ = foo as usize;
}

[Manishearth]

You need to handle both, there's an automatic coercion from FnDef to FnPtr. You'll hit the FnPtr edge for cases where you explicitly cast a variable of type fn(...) into an int.

(FnDef is a zero-sized type that is unique for each function, letting things like .map(func) monomorphize and inline completely)

Testcase is ok.

[phansch]

I added the handling of fn in variable types and function arguments and pushed the commit with the fn_to_numeric_cast lint. I don't have much more time today but will get back to it tomorrow.

[flip1995]

#[warn()] -> #![warn()]

[phansch]

I pushed a first version of the fn_to_numeric_cast_with_truncation lint (which also fixes #2981) and also improved the description of fn_to_numeric_cast (which should close #2980).

I'm not 100% clear on two things:

1. Do the tests for fn_to_numeric_cast_with_truncation make sense? I assume they will fail if the tests are executed on platforms with different pointer sizes.
2. On a platform with 64 bit pointer size, is it safe to cast a function pointer to i64? If not, then I think I'll have to adjust the check.

[flip1995]

There are compile-test flags you can set to ignore tests on specific architectures.

[Manishearth]

The changes look good! I may hold off on merging since I'll be talking about this with someone from Mozilla's legal team tomorrow morning anyway as a sanity-check.

You need to update the test expectations btw.

[phansch]

For some reason (maybe #3198), allow(clippy::fn_to_numeric_cast_with_truncation) does not work inside tests/ui/fn_to_numeric_cast.rs. I will collapse both tests into one for now, which doesn't cover every case but should be good enough for a start.

(I also merged in master, let me know if I should rebase instead)

[Manishearth]

merge is fine. Thanks!

[Manishearth]

Probably should have edited out the do not merge from the title, but that's fine, it was ready to be merged 😄

[matthiaskrgr]

Well, this introduced a couple of dogfood failures and compiler warnings according to travis :/

   Compiling clippy_lints v0.0.212 (/home/travis/build/rust-lang-nursery/rust-clippy/clippy_lints)
warning: constant item is never used: `CLONE`
  --> clippy_lints/src/utils/paths.rs:15:1
   |
15 | pub const CLONE: [&str; 4] = ["core", "clone", "Clone", "clone"];
   | ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
   |
   = note: #[warn(dead_code)] on by default
warning: function is never used: `get_arg_ident`
   --> clippy_lints/src/utils/mod.rs:960:1
    |
960 | pub fn get_arg_ident(pat: &Pat) -> Option<ast::Ident> {
    | ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

failures:
---- [run-pass] run-pass/needless_lifetimes_impl_trait.rs stdout ----
error: compilation failed!
status: exit code: 1
command: "target/debug/clippy-driver" "tests/run-pass/needless_lifetimes_impl_trait.rs" "-L" "/home/travis/build/rust-lang-nursery/rust-clippy/target/debug/test_build_base" "--target=x86_64-unknown-linux-gnu" "-C" "prefer-dynamic" "-o" "/home/travis/build/rust-lang-nursery/rust-clippy/target/debug/test_build_base/needless_lifetimes_impl_trait.stage-id" "-L" "target/debug" "-L" "target/debug/deps" "-Dwarnings" "-L" "/home/travis/build/rust-lang-nursery/rust-clippy/target/debug/test_build_base/needless_lifetimes_impl_trait.stage-id.aux"
stdout:
------------------------------------------
------------------------------------------
stderr:
------------------------------------------
error: explicit lifetimes given in parameter types where they could be elided
  --> tests/run-pass/needless_lifetimes_impl_trait.rs:17:5
   |
17 | /     fn baz<'a>(&'a self) -> impl Foo + 'a {
18 | |         Baz { bar: self }
19 | |     }
   | |_____^
   |
note: lint level defined here
  --> tests/run-pass/needless_lifetimes_impl_trait.rs:3:9
   |
3  | #![deny(clippy::needless_lifetimes)]
   |         ^^^^^^^^^^^^^^^^^^^^^^^^^^
error: aborting due to previous error
------------------------------------------
thread '[run-pass] run-pass/needless_lifetimes_impl_trait.rs' panicked at 'explicit panic', /home/travis/.cargo/registry/src/github.com-1ecc6299db9ec823/compiletest_rs-0.3.14/src/runtest.rs:2550:9
note: Some details are omitted, run with `RUST_BACKTRACE=full` for a verbose backtrace.
stack backtrace:
   0: std::sys::unix::backtrace::tracing::imp::unwind_backtrace
             at libstd/sys/unix/backtrace/tracing/gcc_s.rs:49
   1: std::sys_common::backtrace::print
             at libstd/sys_common/backtrace.rs:71
             at libstd/sys_common/backtrace.rs:59
   2: std::panicking::default_hook::{{closure}}
             at libstd/panicking.rs:211
   3: std::panicking::default_hook
             at libstd/panicking.rs:221
   4: std::panicking::rust_panic_with_hook
             at libstd/panicking.rs:476
   5: std::panicking::begin_panic
             at libstd/panicking.rs:410
   6: compiletest_rs::runtest::ProcRes::fatal
             at /home/travis/.cargo/registry/src/github.com-1ecc6299db9ec823/compiletest_rs-0.3.14/src/runtest.rs:2550
   7: compiletest_rs::runtest::TestCx::fatal_proc_rec
             at /home/travis/.cargo/registry/src/github.com-1ecc6299db9ec823/compiletest_rs-0.3.14/src/runtest.rs:1649
   8: compiletest_rs::runtest::TestCx::run_rpass_test
             at /home/travis/.cargo/registry/src/github.com-1ecc6299db9ec823/compiletest_rs-0.3.14/src/runtest.rs:228
   9: compiletest_rs::runtest::TestCx::run_revision
             at /home/travis/.cargo/registry/src/github.com-1ecc6299db9ec823/compiletest_rs-0.3.14/src/runtest.rs:131
  10: compiletest_rs::runtest::run
             at /home/travis/.cargo/registry/src/github.com-1ecc6299db9ec823/compiletest_rs-0.3.14/src/runtest.rs:81
  11: compiletest_rs::make_test_closure::{{closure}}
             at /home/travis/.cargo/registry/src/github.com-1ecc6299db9ec823/compiletest_rs-0.3.14/src/lib.rs:278
  12: <F as alloc::boxed::FnBox<A>>::call_box
             at liballoc/boxed.rs:672
  13: <F as alloc::boxed::FnBox<A>>::call_box
             at libtest/lib.rs:1461
             at liballoc/boxed.rs:672
  14: __rust_maybe_catch_panic
             at libpanic_unwind/lib.rs:102
failures:
    [run-pass] run-pass/needless_lifetimes_impl_trait.rs
test result: FAILED. 28 passed; 1 failed; 0 ignored; 0 measured; 0 filtered out
test compile_test ... FAILED
failures:
---- compile_test stdout ----
thread 'compile_test' panicked at 'Some tests failed', /home/travis/.cargo/registry/src/github.com-1ecc6299db9ec823/compiletest_rs-0.3.14/src/lib.rs:90:22
note: Some details are omitted, run with `RUST_BACKTRACE=full` for a verbose backtrace.
stack backtrace:
   0: std::sys::unix::backtrace::tracing::imp::unwind_backtrace
             at libstd/sys/unix/backtrace/tracing/gcc_s.rs:49
   1: std::sys_common::backtrace::print
             at libstd/sys_common/backtrace.rs:71
             at libstd/sys_common/backtrace.rs:59
   2: std::panicking::default_hook::{{closure}}
             at libstd/panicking.rs:211
   3: std::panicking::default_hook
             at libstd/panicking.rs:221
   4: std::panicking::rust_panic_with_hook
             at libstd/panicking.rs:476
   5: std::panicking::begin_panic
             at libstd/panicking.rs:410
   6: compiletest_rs::run_tests
             at /home/travis/.cargo/registry/src/github.com-1ecc6299db9ec823/compiletest_rs-0.3.14/src/lib.rs:90
   7: compile_test::run_mode
             at tests/compile-test.rs:68
   8: compile_test::compile_test
             at tests/compile-test.rs:130
   9: compile_test::compile_test::{{closure}}
             at tests/compile-test.rs:128
  10: core::ops::function::FnOnce::call_once
             at libcore/ops/function.rs:238
  11: <F as alloc::boxed::FnBox<A>>::call_box
             at libtest/lib.rs:1468
             at libcore/ops/function.rs:238
             at liballoc/boxed.rs:672
  12: __rust_maybe_catch_panic
             at libpanic_unwind/lib.rs:102
failures:
    compile_test
test result: FAILED. 0 passed; 1 failed; 0 ignored; 0 measured; 0 filtered out
error: test failed, to rerun pass '--test compile-test'

[Manishearth]

That was previously broken, perhaps introduced by a rust compiler change?

[Manishearth]

rust-lang/rust#54741