RFR JDK-8129562: JDK 9 build using boot-jdk classes instead of newly compiled classes

[Magnus Ihse Bursie]

On 2015-08-06 19:08, Jan Lahoda wrote:
> Jon suggested that now, when the EMPTY_BOOTCLASSPATH is used not only 
> for -bootclasspath, it could be renamed to EMPTY_DIR.
>
> An updated webrev with the rename is here:
>
> http://cr.openjdk.java.net/~jlahoda/8129562/webrev.01/
>
> How does this look?
Looks fine to me.

/Magnus
>
> Thanks,
>     Jan
>
> On 6.8.2015 17:33, Tim Bell wrote:
>> Well, OK, since you are already using $(EMPTY_BOOTCLASSPATH) with
>> -bootclasspath
>>
>> Tim
>>
>> On 08/06/15 07:57, Jonathan Gibbons wrote:
>>> Looks good to me, but we need a build-dev Reviewer.
>>>
>>> -- Jon
>>>
>>> On 08/04/2015 02:11 PM, Jan Lahoda wrote:
>>>> Hello,
>>>>
>>>> Any comments on this?
>>>>
>>>> Thanks,
>>>>     Jan
>>>>
>>>> On 27.7.2015 13:49, Jan Lahoda wrote:
>>>>> Hi,
>>>>>
>>>>> Bug:
>>>>> https://bugs.openjdk.java.net/browse/JDK-8129562
>>>>>
>>>>> As part of the fix for JDK-8054717, CompileJavaModules.gmk is now 
>>>>> using
>>>>> an "empty" bootclasspath (classes like java.lang.Object are loaded 
>>>>> from
>>>>> the ordinary classpath as needed). Unfortunately, javac is still 
>>>>> using
>>>>> the default ext and endorsed dirs (if available in the boot JDK), and
>>>>> classes that are in the ext dirs have precedence over the classes 
>>>>> from
>>>>> the classpath. Which may cause compilation problems with the ext dirs
>>>>> contain an older version of a class.
>>>>>
>>>>> The proposal is to make ext and endorsed dirs "empty" as well 
>>>>> (both ext
>>>>> and endorsed dirs will contain one entry, and the entry will be an
>>>>> empty
>>>>> directory):
>>>>> http://cr.openjdk.java.net/~jlahoda/8129562/webrev.00/
>>>>>
>>>>> What do you think?
>>>>>
>>>>> Thanks,
>>>>>      Jan
>>>
>>