[fix] #3928: Make failing the wasm tests actually fail the CI

[DCNick3]

Description

Add a new iroha_wasm_test_runner tool. It mostly duplicates the webassembly-test-runner, but actually returns a non-zero exit code if there are any test failures.

This tool is expected to be installed on the developer's system with cargo install. It's not going to change often, so I don't think there's a risk of it going out-of-sync with the state of the repo.

Also fix the evaluate_expression test, which was never passing before due to a typo.

Linked issue

Closes #3928

Benefits

• Actually makes the CI check that the wasm test

Alternatives

An alternative would be to continue to use the upstream test runner when/if matklad/webassembly-test#1 gets merged, but I wouldn't bet on it.

Checklist

• make sure CI passes

[coveralls]

Pull Request Test Coverage Report for Build 6416151955

• 0 of 46 (0.0%) changed or added relevant lines in 1 file are covered.
• 6351 unchanged lines in 108 files lost coverage.
• Overall coverage decreased (-0.9%) to 58.572%

---

Changes Missing Coverage	Covered Lines	Changed/Added Lines	%
tools/wasm_test_runner/src/main.rs	0	46	0.0%

Files with Coverage Reduction	New Missed Lines	%
cli/src/main.rs	1	0.0%
cli/src/samples.rs	1	0.0%
cli/src/torii/pagination.rs	1	98.9%
config/base/src/runtime_upgrades.rs	1	51.72%
config/src/torii.rs	1	96.67%
core/src/smartcontracts/isi/block.rs	1	87.5%
crypto/src/merkle.rs	1	96.23%
data_model/src/domain.rs	1	48.25%
logger/src/lib.rs	1	95.15%
config/src/lib.rs	2	0.0%

Totals
Change from base Build 5423219773:	-0.9%
Covered Lines:	21698
Relevant Lines:	37045

---

💛 - Coveralls

[mversic]

I see you also opened a PR to webassembly-test-runner repo. Do you think we should use it when your fix is merged?

[DCNick3]

Yeah, I think it's fine to use it when/if it'll get merged. But, AFAIK, matklad is not doing any rust stuff recently, so it might be a while.

[DCNick3]

The iroha_wasm::tests::execute_instruction test fails when CARGO_INCREMENTAL is 0, but passes when it's 1. I am not sure how to even approach debugging this...

[mversic]

@Arjentix can you help Nikita debug this?

[Arjentix]

Will try

[DCNick3]

I have some progress, but not much: it's probably a result of some unsoundness, or a compiler bug, as modifying the code (like inserting a call to a log function) can make the failure disappear. It also seems to occur in release regardless of whether the incremental compilation is used

[DCNick3]

It's a double free caused by the confusion abt whether the function should take ownership.

While debugging this, I've added ad-hoc logging capabilities to the test runner. Not sure if this logger should stay. (It would allow us to move back to the upstream webassembly-test-runner, as my PR was merged)

[DCNick3]

Removed the ad-hoc logging system & made another PR to the upstream (for printing the backtrace of test failures). If it gets merged I'll get rid of the iroha_wasm_test_runner.

[mversic]

Removed the ad-hoc logging system & made another PR to the upstream (for printing the backtrace of test failures). If it gets merged I'll get rid of the iroha_wasm_test_runner.

ok, since this PR is unlikely to conflict with any of the current work we can wait a few days before merging

[6r1d]

LGTM, thanks for a detailed explanation, hopefully we can rebuild the image soon :-)

[DCNick3]

Reverted to hosting the wasm runner tool in monorepo, as the upstream would still be effectively only maintained by us