[foreign-abi] RFR: 8228762: import CallingSequenceBuilder unit tests

[Jorn Vernee]

Looks good!

Jorn

On 2019-08-01 09:34, Nick Gasson wrote:
> Hi,
> 
> I've updated the webrev with the ArgumentClass test removed:
> 
> http://cr.openjdk.java.net/~ngasson/foreign/8228762/webrev.1/
> 
> OK to push?
> 
> Thanks,
> Nick
> 
> 
> On 31/07/2019 22:03, Maurizio Cimadamore wrote:
>> Yeah - that test is very old and it was one of the few tests we had on
>> the ABI when I started working on that.
>> 
>> I think we have much better (and more portable tests) now, and this 
>> can
>> go (perhaps even from foreign branch).
>> 
>> Cheers
>> Maurizio
>> 
>> On 31/07/2019 03:23, Jorn Vernee wrote:
>>> I'm okay with dropping it. ArgumentClassTest actually pre-dates me
>>> being a part of the project :) (I added CallingSequenceTestBase not
>>> too long ago in this area though, when I added the Windows version of
>>> CallingSequenceTest)
>>> 
>>> Jorn
>>> 
>>> On 2019-07-31 10:10, Nick Gasson wrote:
>>>> Hi Maurizio,
>>>> 
>>>>> Looks good, but I'm a bit puzzled by the ArgumentClassTest; in 
>>>>> general
>>>>> the test seems pretty obscure - but there's also the problem that 
>>>>> the
>>>>> test is ultimately SysV specific, so it will only stress the
>>>>> ArgumentClassImpl specific to SysV. Not sure this test is robust 
>>>>> enough
>>>>> (as the other) to make the cut, to be honest.
>>>>> 
>>>>> What do you think?
>>>>> 
>>>> 
>>>> This one got included because I took the whole test/jdk/foreign/abi
>>>> directory from the foreign branch. I agree it's not as useful as the
>>>> main CallingSequenceTest which should test the merge method 
>>>> indirectly
>>>> anyway. I'm happy to take it out of this patch if Jorn agrees too? 
>>>> (I
>>>> think he added it originally.)
>>>> 
>>>> 
>>>> Nick