Ok, I verified with TCK team that there are lots of TCK tests enforcing IAE. So perhaps it's a bad idea to change it now.
Marek
On Jun 15, 2012, at 1:21 PM, Sergey Beryozkin wrote:
> On 15/06/12 11:55, Marek Potociar wrote:
>>
>> On Jun 11, 2012, at 12:31 PM, Sergey Beryozkin wrote:
>>
>>> Hi
>>> On 05/06/12 16:59, Bill Burke wrote:
>>>> Isn't there TCK tests that test that ISE is thrown when null is passed in?
>>>>
>>>
>>> In some cases 'null' is valid (resetting some of the previous state), so it would be good to get some further clarifications on the issue
>>>
>>> NPE is associated with a code bug to me, I often see people saying NPE is always a bug in the code, so I wonder if such a replacement is really necessary
>>
>> The point is that in this case the NPE really indicates a code bug. In fact, the exception thrown now (and declared in API) is IllegalArgumentException (sorry for the confusion) and it' s only purpose is to indicate a code bug too. It's just that by convention, NPE should be used for the null pointer related issues.
>>
> OK, agreed about the convention.
>
> However I can see a lot of tests in our code, some of them derived from TCK based tests, where IllegalArgumentException is checked.
> Replacing it with NPE would not be a big issue I guess, though I find IllegalArgumentException being good enough in this case (null is not a legal argument in this case - but can be a legal one in cases like uriBuilder.replacePath(null)) though it's not as precise as NPE in cases when null is passed but only initialized values are expected.
>
> Cheers, Sergey
>
>> Marek
>>
>>>
>>> Cheers, Sergey
>>>
>>>
>>>> On 6/5/12 11:05 AM, Marek Potociar wrote:
>>>>> Hello Experts,
>>>>> please provide your feedback: http://java.net/jira/browse/JAX_RS_SPEC-131
>>>>>
>>>>> Thanks,
>>>>> Marek
>>>>
>>
>
>
> --
> Sergey Beryozkin
>
> Talend Community Coders
> http://coders.talend.com/
>
> Blog: http://sberyozkin.blogspot.com