Rethinking TypeAnnotation vs Annotation

[Jonathan Gibbons]

I am OK with seeing this idea explored at some point in the anno cleanup 
pipeline.

My expectation is that if you merge TypeCompound and Compound, then we 
will start getting a better feeling of the overall usages of the class, 
and can refactor appropriately.

-- Jon

On 02/06/2014 04:16 PM, Eric McCorkle wrote:
> Could this be done as part of the more general annotations work?
>
> I'm liking Joel's idea of dropping TypeCompound altogether.  Maybe that
> merge should be done now, and we can do the split later?  I think it
> would be a lot easier once the type annotations work is done.
>
> On 02/06/14 17:00, Jonathan Gibbons wrote:
>> Right now, Compound is used for two things, it is the representation of
>> an annotation as an arg value of an enclosing annotation and it is used
>> for "top level" annotation values.  I suspect it is time to split those
>> uses.
>>
>> -- Jon
>>
>> On 02/06/2014 01:48 PM, Werner Dietl wrote:
>>> Hi all,
>>>
>>> an alternative would be to keep Compound only while it is unclear
>>> whether an annotation is a type or declaration annotation and
>>> introduce a new DeclarationCompound class that can be used when only
>>> declaration annotations are expected.
>>> Why not have the type system help us keep things straight?
>>>
>>> cu, WMD.