Test reform

[rrbutani]

what

Tracking issue for cargo t showing all green.

The idea is that you should be able to clone this repo, run nix develop and then cargo t and have all the tests run and pass.

Specifically:

• all the tests should pass
• running the tests should require no extra setup
• there should be no uncaptured stdout/stderr output from the tests
• we should have no flaky tests
• we should measure and report test coverage
• cargo t should run in <10s
• we should clean up test macros and APIs
  • i.e. having single_test, etc. support labels
• we should make some kind of judgement about what adequate testing for each module is/what we want to add
  • prop tests, fuzzing, etc.
• we should decide if we want to expose parts of the test suite to implementors/how to best do this
  • i.e. for the instruction interpreter, perhaps have all the tests reference a macro-defined crate::init() -> impl InstructionInterpreter (or something like that)

steps

The first step is to get all the tests passing.

Doing this on a crate by crate basis, taking the opportunity to do some test related cleanup as we go:

• application-support (no tests, currently)
• baseline-sim ()
• benches ()
  • these are still run as part of cargo t but not actually measured
• device-support ()
• isa ()
• macros (no tests, currently)
• os ()
• shims ()
• test-infrastructure ()
• traits ()