[classlib] do we have any current instances where we've chosen to disagree with the spec or RI?

[Geir Magnusson Jr <ge...@pobox.com>]

If so, lets be sure to log them in JIRA - I created a category for this 
(I forget the name, but it will be obvious...)

geir

Re: [classlib] do we have any current instances where we've chosen to disagree with the spec or RI?

[Geir Magnusson Jr <ge...@pobox.com>]

Tim Ellison wrote:
> Yes we do.  I'm not convinced that logging them in JIRA is the right
> thing -- I'd like to see them a bit closer to the code.

Having them in JIRA doesn't exclude doing anything else, just serves as 
an additional tracking system - a central directory of how we deviate so 
we can keep easy track of the status of any deviation.

> 
> IMHO when we agree on a departure from the spec/RI then the differences
> should be embodied in an API test case.   We should expect the test to
> pass on harmony code, and fail on the RI -- then add that test to the RI
> exclusion list so that all our remaining API tests pass on the RI.  The
> exclusion list will become the repository of differences.  We can
> address each exclusion if/when they are resolved in the RI.
> 

That works, but a set of test cases doesn't capture what I was hoping to 
find in the JIRA, such as a simple way to search, and english-language 
description.  THe JIRA should point to the test case, and a mail archive 
of the discussion of the decision.

geir

> Regards,
> Tim
> 
> Geir Magnusson Jr wrote:
>> If so, lets be sure to log them in JIRA - I created a category for this
>> (I forget the name, but it will be obvious...)
>>
>> geir
>>
> 

Re: [classlib] do we have any current instances where we've chosen to disagree with the spec or RI?

[Tim Ellison <t....@gmail.com>]

Yes we do.  I'm not convinced that logging them in JIRA is the right
thing -- I'd like to see them a bit closer to the code.

IMHO when we agree on a departure from the spec/RI then the differences
should be embodied in an API test case.  We should expect the test to
pass on harmony code, and fail on the RI -- then add that test to the RI
exclusion list so that all our remaining API tests pass on the RI.  The
exclusion list will become the repository of differences.  We can
address each exclusion if/when they are resolved in the RI.

Regards,
Tim

Geir Magnusson Jr wrote:
> If so, lets be sure to log them in JIRA - I created a category for this
> (I forget the name, but it will be obvious...)
> 
> geir
> 

-- 

Tim Ellison (t.p.ellison@gmail.com)
IBM Java technology centre, UK.