Future of 'type-annotations/type-annotations' forest

[Jonathan Gibbons]

Werner,

Potentially, yes. We're investigating.

-- Jon

On 05/09/2014 06:15 PM, Werner Dietl wrote:
> Eric,
>
> are you saying that that part of ClassReader is dead code that will
> never run? Shouldn't it be removed then?
>
> cu, WMD.
>
> On Fri, May 9, 2014 at 8:42 PM, Eric McCorkle <eric.mccorkle at oracle.com> wrote:
>> On 05/08/14 16:57, Eric McCorkle wrote:
>>
>>> I am aware of the fix that was pushed into the type annotations forest.
>>>   As we discussed when you developed the fix, I cannot push it into
>>> jdk9-dev as-is, because the development practices of the langtools team
>>> require that any changeset that introduces behavioral changes be
>>> accompanied by a test.
>>>
>>> I will take care of the test this time, so that you can run checkers on
>>> the contents of jdk9.  However, in the future, we ask that patches
>>> submitted against jdk9 conform to the langtools group's development
>>> practices.
>> I looked into this more.  I'm not at all certain how you even get this
>> problem to show up.  The only type annotations with local variable info
>> are LOCAL_VARIABLE or RESOURCE_VARIABLE type annotations, both of which
>> javac only ever generates on code attributes.  Also, javac never *reads*
>> annotations on code attributes.  Therefore, it is not really possible to
>> write a test.  It also suggests you're running into funny class files in
>> checkers somehow.
>>
>> On the other hand, all this means we can give the bug the noreg-hard
>> tag, and send it through the review process.  So it should be in jdk9 soon.
>
>