RFR: (S) 7152791 wbapi tests fail on cygwin

[Bengt Rutisson]

Mikael,

Thanks for adding the comments.

I'll help you push this. The job is in the hotspotwest queue now.

Bengt

On 2012-03-21 09:17, Mikael Gerdin wrote:
> Thanks Bengt and John,
>
> I added a comment to the MKS/cygwin detection logic to make the MKS 
> dectection more clear.
>
> Webrev including comment change:
> http://cr.openjdk.java.net/~mgerdin/7152791/webrev.1/
>
> I'll also need someone to sponsor this change.
>
> /Mikael
>
> On 2012-03-21 08:34, Bengt Rutisson wrote:
>>
>> Mikael,
>>
>> Looks good!
>>
>> Bengt
>>
>> On 2012-03-14 14:38, Mikael Gerdin wrote:
>>> Hi all,
>>>
>>> Buglink:
>>> http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=7152791
>>> Webrev:
>>> http://cr.openjdk.java.net/~mgerdin/7152791/webrev.0/
>>>
>>> Summary: test/Makefile does not handle cygwin's response to "uname -s"
>>> and cygwin-style paths.
>>> This caused the test "test/sanity/WBApi.java" to fail because it was
>>> called through test/Makefile.
>>>
>>> To run the test through the makefile one would use a command line like:
>>> cd test; make JT_HOME=$MY_JTREG PRODUCT_HOME=$MY_JDK wbapitest
>>>
>>> I modified the Windows detection logic a bit and introduced a
>>> "GETMIXEDPATH", similar to how the JDK repository handles the same
>>> issue. I did not modify the "packtest" or "jtreg_tests" targets since
>>> they aren't executed as far as I know.
>>>
>>> Verified the fix by running the makefile target in cygwin manually and
>>> by running JPRT both with the default "release" and with "-release 
>>> jdk8"
>>>
>>> Thanks
>>> /Mikael
>>