compile-time error handling

[smcclellan]

There are several places in the code where an InternalError is being thrown, wrapping an Unchecked. This should use the proper underlying error to match the interpreter instead.

[csgui]

Adding more details.

Consider the following expression, that should throw an error: (let ((map 2)) (+ map map)).

Such expressions cause the interpreter to throw an Unchecked(NameAlreadyUsed("map")) while the compiler is handling this error as a GeneratorError::InternalError(…).

The compiler should reflect how the interpreter handles the error. Ideally working with these errors from stacks-core.

The generic GeneratorError::InternalError(…) should really just be used for cases that we don't expect to show up.

[Acaccia]

@csgui @smcclellan I don’t understand this issue. Why does it exist, when #386 exists and was closed because we already have InternalErrors ?

[csgui]

@Acaccia I apologize that I closed issue #386 before an investigation into the need for an immediate work on compile-time errors. I should have reopened it instead. Let's consider this one for compilation-time error handling.