Simplify event assertion with predicate-based check

[raymondkfcheung]

A follow-up PR to simplify event assertions by introducing contains_event, allowing event checks without needing exact field matches. This reduces redundancy and makes tests more flexible.

Partially addresses #6119 by providing an alternative way to assert events.

Reference: PR #7594 - Discussion

[raymondkfcheung]

/cmd prdoc --audience runtime_dev --bump patch

[paritytech-workflow-stopper]

All GitHub workflows were cancelled due to failure one of the required jobs.
Failed workflow url: https://github.com/paritytech/polkadot-sdk/actions/runs/13549250284
Failed job name: fmt

[raymondkfcheung]

/cmd fmt

[bkontur]

I've never used this, but from what I’m reading, #[track_caller] is used for functions that can panic or assert, allowing you to see the line where they were called rather than the line where the panic or assertion occurred.
So, this contains_event returns bool, so I think it shouldn't fail.

Suggested change

#[track_caller]

If we use this all the time as assert!(System::contains_event(, maybe we could just change existing:

-  pub fn assert_has_event(event: T::RuntimeEvent)`
+  pub fn assert_has_event(predicate: F)`

[bkchr]

Actually if we look above, we already have assert_has_event. So this can just be used here already and we don't need contains_event.

[raymondkfcheung]

Yes, I agree that we already have a similar function, assert_has_event(). Refactoring it to accept a predicate could be an option. The goal here is to simplify event assertions and provide an alternative way to check for events. Here's a comparison of assert_has_event() vs contains_event():

relay_chain::System::assert_has_event(relay_chain::RuntimeEvent::XcmPallet(
    pallet_xcm::Event::Attempted {
        outcome: Outcome::Complete { used: Weight::from_parts(1_000, 1_000) },
    },
));
		
assert!(relay_chain::System::contains_event(|event| {
    matches!(
        event,
        relay_chain::RuntimeEvent::XcmPallet(pallet_xcm::Event::Attempted { .. })
    )
}));

My original intent was to keep this as a lightweight PR. Refactoring assert_has_event() would require more extensive changes. If that's the preferred approach, I’ll skip adding contains_event() to frame and instead refactor it as a local helper within the XCM scope.

[raymondkfcheung]

@bkchr Removed contains_event() from frame, please check.

[bkchr]

Actually if we look above, we already have assert_has_event. So this can just be used here already and we don't need contains_event.

[serban300]

Nit: I would rename this function since it's not clear who should contain the event.

Also could we make it a macro ? I think it would be easier to use. At least for the use cases that I see in this PR. Something like assert_system_event!(event).

[raymondkfcheung]

I guess: impl_assert_events_helpers_for_relay_chain and impl_assert_events_helpers_for_parachain.

[serban300]

No, no, I mean something like:

macro_rules! assert_system_event {
	( $expected_event:pat ) => {
		relay_chain::System::events().iter().any(|e| matches!(e.event, $expected_event))
	};
}

And then we can just do

assert_system_event!(relay_chain::RuntimeEvent::XcmPallet(pallet_xcm::Event::Attempted { .. }))

Or maybe we should provide relay_chain::System also as a parameter, but anyway, something like that.

[raymondkfcheung]

Renamed

[raymondkfcheung]

Okay, will do.

[serban300]

Sorry, my code is incomplete. You also have to check that relay_chain::System::events().iter().any(|e| matches!(e.event, $expected_event)) actually returns true

[raymondkfcheung]

Instead of assert_system_event, I changed the logic a bit to system_contains_event, which returns a bool. This allows tests to check for both the presence and absence of events. It also works for both parachain and relay chain events.

[serban300]

Looks good ! I just left one more comment, but it's mostly a nit.

[serban300]

RuntimeEvent is hardcoded and it's the same for all runtimes. But I think System is not. The runtimes can use any name when adding the system pallet:

construct_runtime! {
	pub enum Runtime
	{
		System: frame_system,
        ...
     }
}

So if we need to use this method in more tests in the future, we might need to add a parameter to the macro for providing the name of the system pallet. But here it works like this.

Also I would rename module to runtime

[raymondkfcheung]

Renamed. Will change later if we need more parameters.