new tests in type-annotations repo

[Jonathan Gibbons]

On 08/30/2013 11:33 AM, Steve Sides wrote:
> I may be wrong but I think these were meant to test type annotations. 
> At the time these were written I think what we now know as TYPE_USE 
> was assumed to be part of the default set of elements which could be 
> annotated, and tests were to ensure that type annotations 
> implementation did not cause annotations to be accepted where they 
> shouldn't.  That obviously has since changed and now type annotations 
> must be explicitly targeted to TYPE_USE.  So.... (I assumed) that 
> these tests should check that type annotations are not accepted where 
> they shouldn't be.
> I think in keeping with what these were meant to test and testing type 
> annotations, they should be okay.
>
> Some tests could additionally use regular annotations, but some of the 
> test cases are such that it may not be so clear that the type 
> annotation alone is causing the errors and we'd need a second test.  I 
> checked the neg test for annotations (annotations/neg) and there are 
> just a few the check misplacement of annotations, so that argument can 
> be made to increase that type of testing.
>
> Also, the changeset was made against jdk8-tl and the tests pass there, 
> so it could go there also.
>
> -steve 



I still want to go through these tests.  It is not really enough that 
the tests should pass; they should be seen to pass for the right reason.

I could certainly imagine more tests in this area, but it seemed to me 
the goal of this changeset is to fix existing tests, and not provide a 
whole lot more.

I think it would be better to get this changeset approved for 
tl/langtools if possible.

-- Jon