Confusing error message

[Per Bothner]

On 04/21/2011 03:49 PM, Charles Turner wrote:
> On 17/04/2011 08:04, Per Bothner wrote:
>>
>>>
>>> Type checking seems to be a mixed bag in Kawa, most of the time the type
>>> errors seem to emanate from a ClassCastException, which seems quite
>>> elegant.
>>> I'll have to look into why cdr doesn't do it like this, but
>>> when I find out why, would it be okay to try and make it use the
>>> ClassCastException as well?

Well, WrongType has more structure, which is preferable.
Perhaps it might make sense to have WrongType extend ClassCastException.

>>
>> I think so.
>
> I had a look at this today, but the Kawa source got the better of me.
> For some stupid reason, I figured that the ClassCastException would be
> thrown with a custom detail string, how wrong I was.
> For the contrived example of evaluating (cddr '()) [one that throws
> ClassCastExceptions currently] the exception is thrown from
> ModuleMethod.apply1 in the expression,
>
> return module.apply1(this, arg1);
>
> I was looking at this in a debugger and noticed that module is of type
> "lists" which from a quick inspection doesn't have an apply1 method
> (it's a scheme file), I'm struggling to understand how this is working
> at all!

If it was easy, anyone could do it.

The apply1 is the one defined in ModuleBody.  lists.scm is compiled to a
class lists - which extends ModuleBody.

The apply1 method is generated by 
Compilation#generateApplyMethodsWithoutContext.

A ClassCastException is caught and turned into a WrongType by the
CheckedTarget#emitCheckedCoerce.

This is pretty hairy stuff, since it depends on understanding bytecode
and how it is generated.  If you figure that out, you'll be in rare
company!
-- 
	--Per Bothner
per@bothner.com   http://per.bothner.com/