hg: jigsaw/jigsaw/hotspot: Summary: Allow passing in a non-null module loader to translate to boot loader.

[Karen Kinnear]

A couple more details on that.

1) As Alan pointed out, for existing applications, they should see no difference.
2) When running with modules, this change is to accept a non-null loader for the base module. 
There are classes which today load with the vm boot loader, i.e. a null class loader, which are not
in the base module and so they would already return a non-null loader. e.g. xml, corba, ...

thanks,
Karen

On May 23, 2012, at 11:52 AM, Mandy Chung wrote:

> On 5/23/2012 5:47 AM, Paul Sandoz wrote:
>> Hi Karen,
>> 
>> Is is correct to assume that something else needs to happen to make:
>> 
>>   Object.class.getClassLoader() != null
>> 
>> e.g. a tweak to the Class.getClassLoader() method?
> 
> That's right.   With Karen's change, the VM is able to accept non-null loader for the base module.  In the current jigsaw/jdk prototype, it can only take null loader if it requests a class in the base module.
> 
> My next step is to experiment with Class.getClassLoader() returning non-null and evaluate its implication.
> 
> Mandy
> 
>> Paul.
>> 
>> On May 23, 2012, at 2:48 AM, karen.kinnear at oracle.com wrote:
>> 
>>> Changeset: c51a2d4dada0
>>> Author:    acorn
>>> Date:      2012-05-18 13:34 -0400
>>> URL:       http://hg.openjdk.java.net/jigsaw/jigsaw/hotspot/rev/c51a2d4dada0
>>> 
>>> Summary: Allow passing in a non-null module loader to translate to boot loader.
>>> Reviewed-by: mchung
>>> 
>>> ! make/linux/makefiles/mapfile-vers-debug
>>> ! make/linux/makefiles/mapfile-vers-product
>>> ! make/solaris/makefiles/mapfile-vers
>>> ! src/share/vm/prims/jni.cpp
>>> ! src/share/vm/prims/jvm.cpp
>>> ! src/share/vm/prims/jvm.h
>>>