This commit refactors the tests/wast.rs test suite which runs all of the upstream spec tests as *.wast files as well as our own misc_testsuite which has its own suite of *.wast files. Previously the set of wasm features active for each test was a sort of random mishmash and convoluted set of conditionals which was updated and edited over time as upstream proposal test suites evolved. This was then mirrored into our own conventions for misc_testsuite as well. Overall though this has a number of downsides I'm trying to fix here:
The calculation of what features are enabled is quite complicated and effectively a random mishmash of || conditionals with hierarchies that don't make any sense beyond "this is just required to get things to pass".
There is no means of per-test configuration. For example canonicalize-nans.wast had hardcoded logic in tests/wast.rs that it needed a different setting turned on in Config.
There was no easy means to write tests for Wasmtime which take a union of a number of proposals together without having lots of sub-folders that may not make sense.
Tests that require a particular proposal had to have duplicate logic for Winch as it doesn't support the full suite of features of all proposals that Cranelift does.
The new system implemented in this commit takes a leaf out of the disas tests. There is a new TestConfig structure in the tests/wast.rs harness which is decoded from each test (leading ;;! comments) which enables specifying, in each test, what's required. This encompasses many wasm proposals but additionally captures other behavior like nan-canonicalization. This means that all test files in misc_testsuite/**/*.wast are now manually annotated with what wasm features they require and what's needed to run. This makes per-test configuration much easier, per-config-setting much easier, and blanket ignore-by-proposal for Winch much easier as well.
For spec tests we can't modify the contents of the upstream *.wast files. To handle this they're handled specially where TestConfig is manually created and manipulated for each spec proposal and the main test suite itself. This enables per-proposal configuration that doesn't leak into any others and makes it more obvious what proposals are doing what.
This commit refactors the
tests/wast.rs
test suite which runs all of the upstream spec tests as*.wast
files as well as our ownmisc_testsuite
which has its own suite of*.wast
files. Previously the set of wasm features active for each test was a sort of random mishmash and convoluted set of conditionals which was updated and edited over time as upstream proposal test suites evolved. This was then mirrored into our own conventions formisc_testsuite
as well. Overall though this has a number of downsides I'm trying to fix here:The calculation of what features are enabled is quite complicated and effectively a random mishmash of
||
conditionals with hierarchies that don't make any sense beyond "this is just required to get things to pass".There is no means of per-test configuration. For example
canonicalize-nans.wast
had hardcoded logic intests/wast.rs
that it needed a different setting turned on inConfig
.There was no easy means to write tests for Wasmtime which take a union of a number of proposals together without having lots of sub-folders that may not make sense.
Tests that require a particular proposal had to have duplicate logic for Winch as it doesn't support the full suite of features of all proposals that Cranelift does.
The new system implemented in this commit takes a leaf out of the
disas
tests. There is a newTestConfig
structure in thetests/wast.rs
harness which is decoded from each test (leading;;!
comments) which enables specifying, in each test, what's required. This encompasses many wasm proposals but additionally captures other behavior like nan-canonicalization. This means that all test files inmisc_testsuite/**/*.wast
are now manually annotated with what wasm features they require and what's needed to run. This makes per-test configuration much easier, per-config-setting much easier, and blanket ignore-by-proposal for Winch much easier as well.For spec tests we can't modify the contents of the upstream
*.wast
files. To handle this they're handled specially whereTestConfig
is manually created and manipulated for each spec proposal and the main test suite itself. This enables per-proposal configuration that doesn't leak into any others and makes it more obvious what proposals are doing what.