fix: wire-reform tests

[jodarove]

Details

1. fix: invalid syntax error on invalid wire param value:

When a wire parameter has invalid value (eg: foo..bar, foo.bar[3]) it should (ideally) resolve in a compilation error during validation phase.

This is not possible due that platform components may have a param definition which is invalid but passes compilation, and throwing at compile time would break such components.

In such cases where the param does not have proper notation, the config generated will use the bracket notation to match the current behavior (that most likely end up resolving that param as undefined).

2. test for wire reactive parameters

3. tests for registering wire adapter

4. fix: add backward compatibility when adding a config listener.

5. General tests for wire adapter.

Pending:

Fix tests that are failing by adding the logic into the wire-reform

[caridy]

I don't fully understand what this is.

[jodarove]

@caridy without this fix, the compiler will transform and input like this one: @wire(FooAdapter, { bar: '$a.0b' }) into:

   //..
  config: function ($cmp) {
     let v1 = $cmp.a;
     return {
        bar: v1 != null ? v1.0b : undefined
       };
  }
  //..

and 0b is not a valid identifier, therefore it will result in an invalid syntax.

This fix, transforms it into:

   //..
  config: function ($cmp) {
     let v1 = $cmp['a'];
     return {
        bar: v1 != null ? v1['0b'] : undefined
       };
  }
  //..

in order to preserve backward compatibility.

Ideally, this (and other similar cases) should result in a compilation error, but throwing at compile time for existing components (that weren't throwing) is not an option now.

[caridy]

same here.

[jodarove]

memberExprPropertyGen simply alternates the generation between dot notation (on well formed parameters) and bracket notation (malformed parameters)

[caridy]

any particular reason to freeze these? or that's just to prevent them from poisoning the behavior? in the old code, they never had access to this, so, we didn't protect it. just curious.

[jodarove]

just to prevent poisoning.

[caridy]

you should probably freeze it on the next line after declaration, and make AdapterClass an actual class to it retains the proper name instead of an anonymous class

[jodarove]

makes sense, fixing it.

[caridy]

LGTM, mostly comments about things that I don't understand, but the rest is looking good.

[jodarove]

@caridy i didn't fully understood this test, therefore i leaved for you to review.