Add autogenerated Bolero harnesses

[cvick32]

Description of changes:

Adds 58 Bolero harnesses and all necessary TypeGenerator derivations. These harnesses were automatically generated by a tool currently under development by the Kani team.

Summary of running cargo test listed below.

failures:
    decoder::checked_range::bolero_harnesses::bolero_test_25_len
    encoder::estimator::bolero_harnesses::bolero_test_29_write_repeated

test result: FAILED. 56 passed; 2 failed; 0 ignored; 0 measured;

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.

[camshaft]

I guess you're pulling the variable names from the function declaration but I wonder if we could automatically remove the _ prefix? Not a huge deal, just a minor nit.

[camshaft]

Also I'm assuming this panics when callee.len + count exceeds usize::MAX. With our current implementation, the Encoder impls are expected to panic on extreme values, since they don't return Result anywhere. Not sure what's the best way to capture that - do we filter out inputs in the harness that we expect to panic?

[cvick32]

Not exactly sure how to filter out the inputs. The code below still errors out with the input:

Input: 
(
    EncoderLenEstimator {
        capacity: 0,
        len: 9151314442816847872,
    },
    9295429630892703744,
    0,
)1

which I figured would be filtered by the harness.

#[test]
    #[cfg_attr(kani, kani::proof)]
    fn bolero_test_29_write_repeated() {
        bolero::check!().with_type().cloned().for_each(
            |(mut callee, count, value): (EncoderLenEstimator, usize, u8)| {
                if callee.len + count > usize::MAX {
                  None
                } else {
                  Some(callee.write_repeated(count, value))
                }
            },
        );
    }

[camshaft]

Returning None from for_each will result in the harness failing. You need to use filter instead.

[camshaft]

See impl IntoResult for Option<T>

[camshaft]

you'd need to use the tuple, rather than separate arguments (similar to for_each)

[cvick32]

Tried to add a filter like this:

fn encoder_len_estimator_write_repeated() {
        bolero::check!()
            .with_type()
            .cloned()
            .filter(|(callee, count, _): &(EncoderLenEstimator, usize, u8)| {
                (callee.len + count) < usize::MAX
            })
            .for_each(
                |(mut callee, count, value): (EncoderLenEstimator, usize, u8)| {
                    Some(callee.write_repeated(count, value))
                },
            );
    }

I still get an error in this test but now bolero isn't giving an input that causes the error. Instead I get this:

---- encoder::estimator::tests::encoder_len_estimator_write_repeated stdout ----
thread 'encoder::estimator::tests::encoder_len_estimator_write_repeated' panicked at 'attempt to add with overflow', /rustc/eb26296b556cef10fb713a38f3d16b9886080f26/library/core/src/ops/arith.rs:109:1
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace


failures:
    encoder::estimator::tests::encoder_len_estimator_write_repeated

test result: FAILED. 79 passed; 1 failed; 0 ignored; 0 measured; 0 filtered out; finished in 7.01s

[cvick32]

Oh that filter fails because it overflows in the filter check... This filter should get around that:

filter(|(callee, count, _): &(EncoderLenEstimator, usize, u8)| {
 count <= &(usize::MAX - callee.len)
})

[camshaft]

Looks great! Thanks for making those changes