Review Request : JDK-8031670: remove unneeded -source option from javadoc tests

Joe Darcy joe.darcy at oracle.com
Mon Feb 24 09:12:46 PST 2014


FWIW, I've already filed several bugs against JDK 10 for exactly that 
purpose, recording some of the details of the version update from 8 -> 9 
so we don't have to rediscover all those details going from 9 -> 10.

-Joe

On 02/24/2014 08:04 AM, Neil Toda wrote:
>
> Thanks Vicente.  I like that idea of adding to JBS flagging future 
> required changes.
>
> -neil
>
> On 2/24/2014 5:51 AM, Vicente-Arturo Romero-Zaldivar wrote:
>> Hi Neil,
>>
>> Some comments below.
>>
>> On 21/02/14 21:28, Neil Toda wrote:
>>>
>>>
>>> Hi Vicente..  some additional thoughts/comments below...
>>>
>>> On 2/21/2014 12:53 PM, Vicente-Arturo Romero-Zaldivar wrote:
>>>> Hi Neil,
>>>>
>>>> On 21/02/14 01:25, Neil Toda wrote:
>>>>>
>>>>> There is another test that does a negative check on a feature, 
>>>>> testLambdaFeature.
>>>>> However, it too will stop working when support for 1.7 is removed 
>>>>> from the JDK.
>>>>> Granted that will be several years from now, but by then we'll all 
>>>>> forget and just
>>>>> remove the negative test.
>>>>
>>>> I hope that we won't ever remove negative tests just because they fail.
>>>>
>>>
>>> This test has two parts.  The first checks to see that lambda is 
>>> recognized.
>>> The second part checks that if the version supplied via -source does 
>>> not support
>>> lambda, the case is correctly handled.
>>>
>>> When 1.7 is no longer supported in the JDK, then specifying "-source 
>>> 1.7" will not be
>>> valid.  At that point, all supported versions will support lambda.
>>
>> I would create a task in JBS, something like, update source option in 
>> tests where you can add the comments of what needs to be done and 
>> what tests will need to be modified. Then add a comment in all 
>> affected tests referring to this JBS entry, see for example: 
>> https://bugs.openjdk.java.net/browse/JDK-8006694 and the associated 
>> changeset. This way we won't forget all your findings.
>>
>>
>>
>>>
>>> So I mean that in that case, the second part of the test is no 
>>> longer needed nor valid.
>>>
>>>
>>>>>
>>>>> I think I've made a mistake taking out the ./sourceOption test.  
>>>>> It is better that this
>>>>> test exist and be specifically designed to make sure -source is 
>>>>> working.
>>>>
>>>> Agree.
>>>
>>> So, below when you say you'd like to see the proposed change, you 
>>> are referring to statement
>>> about leaving ./sourceOption in?  I'm just suggesting that my note 
>>> below be added to ./sourceOption
>>> in the form of comment giving guidance for future releases, when 
>>> 1.6, then 1.7 is no longer supported.
>>
>> OK, I agree.
>>
>>>
>>> -neil
>>
>> - Vicente
>>
>>>
>>>>
>>>>>
>>>>> With that in mind, I'll not remove it, and add to it so that we 
>>>>> have a list of features
>>>>> in each release that will fail in the next release. Something like:
>>>>>
>>>>> Release        Feature
>>>>> =======        ===================
>>>>> 1.6            List<String> list = ArrayList<String>();
>>>>> 1.7            List<String> list = ArrayList<>();
>>>>> 1.8            Lambda or default methods
>>>>> 1.9            TBD
>>>>>
>>>>> So for JDK9, the test will specify:
>>>>>     -source 1.6
>>>>>     A.java will contain : List<String> list = ArrayList<>();
>>>>>
>>>>>     and the test will look for failure.
>>>>>
>>>>> In this way, the intent will by conveyed in the test.
>>>>>
>>>>> Does this all sound okay?
>>>>
>>>> I would prefer to see the proposed change, I'm not sure I fully 
>>>> understand what you want to do here.
>>>>
>>>> Thanks,
>>>> Vicente
>>>>>
>>>>> Thanks Joe
>>>>>
>>>>> -neil
>>>>>
>>>>>
>>>>> On 2/19/2014 3:42 PM, Joseph Darcy wrote:
>>>>>> Hi Neil,
>>>>>>
>>>>>> On the removal of
>>>>>>
>>>>>>     test/tools/javadoc/sourceOption/SourceOption.java
>>>>>>
>>>>>> after your patch is there some remaining javadoc test which uses 
>>>>>> the -source option? We probably want to keep one test which uses 
>>>>>> to the option to make sure it is supported. If there is no such 
>>>>>> remaining test, SourceOption.java could be resurrected and make 
>>>>>> to check that a JDK 8 feature is rejected under source 7, etc.
>>>>>>
>>>>>> Otherwise, the removal of the other -source uses looks fine.
>>>>>>
>>>>>> Thanks,
>>>>>>
>>>>>> -Joe
>>>>>>
>>>>>> On 2/19/2014 7:37 AM, Vicente-Arturo Romero-Zaldivar wrote:
>>>>>>> Hi Neil,
>>>>>>>
>>>>>>> there is a typo at test/tools/javadoc/6964914/JavacWarning.java:
>>>>>>>
>>>>>>> "It *amy* be deprecated in JDK8"
>>>>>>>
>>>>>>> at test/com/sun/javadoc/testLambdaFeature/TestLambdaFeature.java:
>>>>>>>
>>>>>>> I would add the note as an independent comment after the
>>>>>>>
>>>>>>> * @test
>>>>>>> * @bug      8004893 8022738
>>>>>>> section not as part of it.
>>>>>>>
>>>>>>> Vicente
>>>>>>>
>>>>>>> On 18/02/14 21:10, Neil Toda wrote:
>>>>>>>>
>>>>>>>> A small set of javadoc tests contain the -source parameter.  In 
>>>>>>>> most cases, the parameter is
>>>>>>>> not required for the test.  In several cases, the -source 
>>>>>>>> parameter is being explicitly tested,
>>>>>>>> but relies on a JDK version that will be removed from JDK9.
>>>>>>>>
>>>>>>>> This changeset removes unnecessary -source specification when 
>>>>>>>> not needed, or changes the
>>>>>>>> test/source-version requested to one that will work in JDK9.
>>>>>>>>
>>>>>>>> http://cr.openjdk.java.net/~darcy/neiltoda/8031670.0/
>>>>>>>>
>>>>>>>> Thanks
>>>>>>>>
>>>>>>>> -neil
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>
>>>>>>
>>>>>
>>>>
>>>
>>
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.openjdk.java.net/pipermail/compiler-dev/attachments/20140224/b09a5869/attachment.html 


More information about the compiler-dev mailing list