jdk8 first build attempt - failed

[David Holmes]

Hi Peter,

On 30/07/2013 12:31 PM, Pete Brunet wrote:
> Looking through the corba section of the make debug log I don't see any
> errors other than two occurrences of this line:
>
> find: ‘/cy/cygdrive/c/Users/Pete/JDK8/jdk8/corba/src/share/classes’:
> No such file or directory
>
> Is that indicative of anything wrong? The prefix and suffix sure look
> strange and that directory nzmd (minus the garbage) does exist.

Yes this is a recently discovered bug affecting some systems. See 
8021430. Workaround (from the bug):

"A little deeper investigation revealed that a call to macro at 
corba/makefiles/BuildCorba.gmk:147 goes to macro "SetupIdlCompilation" 
defined in common/makefiles/IdlCompilation.gmk . The abspath function 
was mangling the path. Commenting out line 87 and 88 (to remove 
abspath's effect on the paths) enabled the build to succeed . "

I just raised the priority of that bug. Not clear what the critical 
environmental factors are.

David
-----


> Pete
>
> On 7/29/13 7:13 PM, Pete Brunet wrote:
>> Thanks Phil,
>>
>> No luck looking through the log so far but will keep looking.
>>
>> BTW, I grep'd for BindingListHolder in all the IDL files under jdk8 with
>> no matches.
>>
>> Pete
>>
>> On 7/29/13 12:36 PM, Phil Race wrote:
>>> On 7/29/2013 10:25 AM, Pete Brunet wrote:
>>>> Thanks Phil, I found
>>>> build\windows-x86-normal-server-fastdebug\corba\gensrc and in that dir
>>>> is one file _the_idls.d with one line
>>>> IDLS_ARE_CREATED=yes
>>>>
>>>> Pete
>>>>
>>> Hmm .. in my case that is there but there are also com & org
>>> directories with
>>> in all about 3Mb of generated sources. So the first failure is really
>>> somewhere
>>> before where you are seeing it reported.
>>>
>>> If you aren't already, build with LOG=debug
>>>
>>> -phil.
>