chore: replacing all generic errors with FuelError

[mvares]

• Closes Replace all occurrences of Error to FuelError #2001

[danielbate]

@msensys does this pick up all occurences?

[petertonysmith94]

@msensys does this pick up all occurences?

It believe not - hence the "Relates to"?

[danielbate]

Sorry @msensys was just trying to validate why the validate changeset CI stage hadn't run so converted to draft

[mvares]

@msensys does this pick up all occurences?

Maybe not. I will send some commits to pick up all occurrences.

[nedsalk]

Thanks for the work!

To fully wrap this up, can you please go to the tests that are expecting these errors and use the expectToThrowFuelError utility?

For example, for this error, there exists this test. You'd change the test to look like this:

it('throws when setting configurable but predicate has none', async () => {
  await expectToThrowFuelError(
    () =>
      new Predicate({
        bytecode: predicateBytesTrue,
        abi: predicateAbiTrue,
        provider: wallet.provider,
        inputData: ['NADA'],
        configurableConstants: {
          constant: 'NADA',
        },
      }),
    new FuelError(
      ErrorCode.INVALID_CONFIGURABLE_CONSTANTS,
      'Predicate has no configurable constants to be set'
    )
  );
});

You can do a repo-wide search for the expected error message on each error you changed and there should be a corresponding test which you'd change as shown above. If a test doesn't exist for the error, you're welcome to add it.

[maschad]

@msensys I see that this PR still needs some work as there are quite a few places we still throw Error as opposed to FuelError (at least in 41 files). I'm going to move this PR to draft and you can re-open it to review once it's ready.

[arboleya]

@msensys I see that this PR still needs some work as there are quite a few places we still throw Error as opposed to FuelError (at least in 41 files). I'm going to move this PR to draft and you can re-open it to review once it's ready.

Note that we should use the UNKNOWN code for unmapped errors.

However, to apply this, you'll need to wait until we merge this:

• chore: add UNKNOWN error code #2562

cc @maschad

[arboleya]

Note that we should use the UNKNOWN code for unmapped errors.

Also, UNKNOWN errors must contain the rawError (see comment).

[petertonysmith94]

@mvares did you need any assistance with this one?

[mvares]

@mvares did you need any assistance with this one?

Some Error class in code are related to tests. Should I change it too?

[petertonysmith94]

@mvares did you need any assistance with this one?

Some Error class in code are related to tests. Should I change it too?

@mvares You can leave Error class usage within test files.

[nedsalk]

Thanks for the work!

[nedsalk]

Oops, didn't see that it's in draft 😅 I'll revoke the approval but, regardless, the PR seems good.

[maschad]

@mvares do you need any further guidance here? I can see doing a rudimentary search in my editor there are still quite a few places where FuelError needs to be integrated.

[mvares]

@maschad, sorry for about that. Currently, other issues are on my radar. I think it’s better I unassign and to close this PR.

[petertonysmith94]

There is surely value in the PR in it's current state - I'm for merging as is, and not closing the issue.

[arboleya]

I couldn't agree more.

I'm afraid we can't restore branches on forks tho.

[maschad]

That's a good point @petertonysmith94 , @mvares if you have the branch locally you could restore it and re-open a PR, otherwise you could recreate the changes on a single commit. If you re-open the PR I suppose instead of using the Closes keyword you could use the Related keyword.

[mvares]

I believe I still have the branch locally. I will check it out.