refactor implementation lookup #146
Conversation
| }); | ||
| prototypeMethodNames.forEach (function(_name) { | ||
| typeClass.methods[_name] = function(x) { | ||
| return (getPrototypeMethod (_name) (x)).bind (x); |
There was a problem hiding this comment.
This is the only place that Function#bind is invoked. At the other getPrototypeMethod call site all we want to know is whether the method exists.
There was a problem hiding this comment.
Yeah, see also 4c9ddd7, where I split the function into two. One for checking, one for binding.
| return x != null && | ||
| getStaticMethod (_name) (x.constructor) != null; |
There was a problem hiding this comment.
The test suite passes without the x != null guard. All of this library's type classes that require static methods happen to depend on type classes that require prototype methods that Null and Undefined cannot provide.
These steps demonstrate that x.constructor should be guarded:
- Remove the
x != nullguard. - Remove the
Semigroupoiddependency fromCategory. - Evaluate
Z.Category.test (null).
Since users can define their own type classes via Z.TypeClass, it is necessary to include the guard.
There was a problem hiding this comment.
users can define their own type classes [which necessitate inclusion of] the guard
That sounds like it might warrant a test.
This pull request removes the awful
funcPathfunction, and splitsgetBoundMethodintogetStaticMethodandgetPrototypeMethod(which can be top-level functions as they do not referencerequirements).Aldwin and I have made changes to these several times over the years. 😂