Disable system calls from xcm & Refactoring emulated tests

[yrong]

• Disable the agent/channel operation from xcm(not in the scope of the next launch)
• Refactoring emulated tests with inbound fixtures
• More tests and checks

Requires: Snowfork/polkadot-sdk#108

[codecov]

Codecov Report

Attention: 6 lines in your changes are missing coverage. Please review.

Comparison is base (ec4df8e) 49.74% compared to head (69d632a) 50.05%.
Report is 1 commits behind head on main.

❗ Current head 69d632a differs from pull request most recent head 77b79b7. Consider uploading reports for the commit 77b79b7 to get more accurate results

Files	Patch %	Lines
...xtures/src/register_token_with_insufficient_fee.rs	0.00%	3 Missing ⚠️
...inbound-queue/fixtures/src/send_token_to_penpal.rs	0.00%	3 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1115      +/-   ##
==========================================
+ Coverage   49.74%   50.05%   +0.31%     
==========================================
  Files          61       63       +2     
  Lines        3701     3736      +35     
  Branches       72       72              
==========================================
+ Hits         1841     1870      +29     
- Misses       1843     1849       +6     
  Partials       17       17              

Flag	Coverage Δ
rust	46.43% <57.14%> (+0.38%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[alistair-singh]

+1

[vgeddes]

It seems we have now have this pattern in the emulated tests where we call these three functions in order:

• refund_relayer
• do_convert
• send_xcm

However this is problematic as internal implementation details are now leaking into the tests. It makes the tests more brittle, and less useful. The tests should treat submit as a black box.

I would rather refactor the submit extrinsic, so that it calls the following function, where verification is optional.

fn process_message(message: &Message, verify: bool) -> DispatchResult

And then the simulated tests can call process_message(msg, false)

[yrong]

Notice @claravanstaden is making InboundQueueFixture auto-generated in #1119, then I would prefer to reuse it for the emulated test here. So there is no need to add the process_message(msg, false) function and just test the original submit function with the verification included.

The problem is that InboundQueueTest is for benchmark only and we want to use it also in unit/emulated tests. We did make #1013 for that purpose though closed it for some other high-priority PRs.

So maybe it's time to reopen it. Meanwhile In #1119 we also take it into consideration when generating InboundQueueFixture.

@vgeddes @claravanstaden WDYT?

[claravanstaden]

I made a comment here: #1119 I would also like to use https://github.com/Snowfork/snowbridge/tree/main/parachain/pallets/ethereum-client/tests/fixtures in the runtime tests. Maybe worthwhile to split test fixtures out to a new crate.

[vgeddes]

Cool, ok, this all makes sense - lets split the fixture out into a separate crate - something like pallets/inbound-queue/fixtures.

[yrong]

Addressed in #1125

[vgeddes]

Ok now that we have the fixtures, its no longer necessary to have this refactor_relayer function. @yrong can you revert it?

Also helps reduce the amount of code which needs to be re-audited.

[yrong]

Revert in 4f9f56e