Re: [abc] test run results

Ondrej Lhotak wrote:

>On Fri, Sep 24, 2004 at 04:10:05PM +0200, Aske Simon Christensen wrote:
>
>
>>>724: new - declare class implements class
>>>
>>>
>>ajc fails this one. <class> implements <class> is not legal according to this
>>test case, but ajc accepts it. Which one should we follow - ajc or the test?
>>
>>I think <class> implements <class> does not make sense...
>>
>>
>
>I agree that class implements class does not make sense, and I wouldn't
>oppose rejecting it. However, atrack does this in at least one instance.
>
>So, we need to make a decision whether we want to compile atrack as it
>is, or whether we want to propose a set of patches needed to compile
>atrack with abc.
>
>It's possible that the atrack people will accept our patches into their
>code base, if we can convince them that they make the code clearer. In a
>case like this, I can't imagine why they would oppose a patch that
>changes class implements class into class extends class.
>
>
Surely if we're aiming for ajc's "'implements' and 'extends' are
interchangeable in 'declare parents'", the thing to do would be to
change <class> implements <class> into <class> extends <class> internally?

Or does that have an associated set of pitfalls?

- P

>
>
>>>Gained:
>>>991: abctests/bugs - declare parents causes interface to implement itself
>>>
>>>
>>When an interface extends itself, it is now an error. I will change it to be a
>>noop, and only be an error if it extends a proper subinterface. That should
>>take care of the atrack case.
>>
>>
>
>That would be great.
>
>Ondrej
>
>
>
>>-Aske
>>
>>
>>
>>
>
>
>
>
>
Received on Fri Sep 24 15:34:41 2004