fix(transaction-controller): Return global ethQuery when !isMultichainEnabled

[legobeat]

Explanation

There is a test stating that "MultichainTrackingHelper.getEthQuery should always return ethQuery for global provider when !isMultichainEnabled`".

It was not, but the test was passing anyway due to the mock being hardcoded to do so. This fixes that.

Also refactors tests to actually test for it (separately in #4391).

---

(Aside: I'm not sure why there even is a distinction between "global provider" and "chain-specific providers" in the first place - seems like there should never be a reason for both to exist simultaneously?) Answer: This is transitory and the "global provider" us due to be refactored away.

References

Related

• testfix(transaction-controller): Refactor provider setup to provide correct providers and tracker #4391

Blocking

• feat: retrieve global provider and block tracker dynamically #4359
• deps(transaction-controller): @metamask/nonce-tracker@^5.0.0->^6.0.0 #4327
• chore(deps): bump @metamask/nonce-tracker from 5.0.0 to 6.0.0 #4329

Changelog

@metamask/transaction-controller

• FIXED: MultichainTrackingHelper.getEthQuery now returns global ethQuery with

Checklist

• I've updated the test suite for new or updated code as appropriate
• I've updated documentation (JSDoc, Markdown, etc.) for new or updated code as appropriate
• I've highlighted breaking changes using the "BREAKING" category above as appropriate

[mcmire]

Makes sense.

[legobeat]

(Conflict resolution)

[dbrans]

Also refactors tests to actually test for it (separately in #4391).

Hi @legobeat, I couldn’t find the test for !this.#isMultichainEnabled in #4391. It’s usually clearer to include tests in the same PR, especially for future reference.

[legobeat]

Also refactors tests to actually test for it (separately in #4391).

Hi @legobeat, I couldn’t find the test for !this.#isMultichainEnabled in #4391. It’s usually clearer to include tests in the same PR, especially for future reference.

The existing test is here.

See also the multichainTrackerHelperMock, which always uses and returns the same provider unconditionally for getEthQuery and getProvider.

[legobeat]

The test-related changes have now been separated into #4391. This PR now contains the fix in isolation.