harmony-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Paulex Yang <paulex.y...@gmail.com>
Subject Re: [classlib][NIO|VMI]JNI 1.4 enhancement on ByteBuffer
Date Wed, 14 Jun 2006 03:07:21 GMT
Jimmy, Jing Lv wrote:
> Paulex Yang wrote:
>> There is some enhancement on JNI spec in JDK 1.4[1], and three 
>> methods are related to java.nio.ByteBuffer.
>>
>>    * |NewDirectByteBuffer|
>>      
>> <http://java.sun.com/j2se/1.4.2/docs/guide/jni/jni-14.html#NewDirectByteBuffer>

>>
>>
>>    * |GetDirectBufferAddress|
>>      
>> <http://java.sun.com/j2se/1.4.2/docs/guide/jni/jni-14.html#GetDirectBufferAddress>

>>
>>
>>    * |GetDirectBufferCapacity|
>>      
>> <http://java.sun.com/j2se/1.4.2/docs/guide/jni/jni-14.html#GetDirectBufferCapacity>

>>
>>
>> Because these methods are actually classlib dependent and JNI 
>> implementation must know some details of ByteBuffer implementation, 
>> current IBM VME hasn't them implemented, and seems DRLVM doesn't 
>> implemented thoroughly(please correct me if I made mistake here, 
>> seems DRLVM tries to get some non-api method/field of ByteBuffer, and 
>> if fails, it return NULL or -1 as JNI spec says). And I have no idea 
>> how Sable/JCHEVM/BootJVM deals with this issue yet.(anyone kindly let 
>> me know?)
>>
>> I propose to provide the implementation in NIO component, and I raise 
>> Harmony-578 for it. The idea is: export these three methods in NIO 
>> module as hynio.dll(.so), which is loaded by Harmony launcher, and 
>> add these methods to VMI in some way, so that the VM vendor(i.e., JNI 
>> implementation vendor) can add these methods to JNI function table.
>> Other choices I can imagine now include:
>> 1. Add related direct buffers class to kernel class, so that the VM 
>> vendor can implement it as well as the JNI methods. IMO this is not 
>> good choice because buffers are actually VM independent, it's not 
>> reasonable to let VM vendor to implement these classes.
>>
>
> By reading the spec, it seems RI prefer this way, take direct-buffer 
> as kernel class ,like class String(Though maybe it is hard to tell 
> "kernel" and "normal" classes in RI's implementation, they're always 
> together there :) ).
>
> And in Harmony, there's an interface named "DirectBuffer" 
> (o.a.h.nio.internal), abstract class Buffer(java.nio) and an 
> implementation class "ReadWriteDirectByteBuffer" (java.nio),which 
> contains fields and methods for JNI methods. So an easy way may be: 
> take these "as" kernel classes, and get Address from 
> DirectBuffer.getBaseAddress(), get Capacity from Buffer.capacity, and 
> new a ReadWriteDirectByteBuffer as basic direct buffer in three JNI 
> methods.
I think the kernel class means the classes which heavily depends on VM 
implementation, but the buffer is another story, it is the JNI actually 
depends on Classlib implementation, so instead of put buffers into 
kernel, I prefer to pull the three JNI methods out of VM into classlib.
>
>> 2. Provides some utility methods in o.a.h.nio, add these methods to 
>> VMI, so that VM vendor can get inside knowledge on the direct buffers 
>> by these utilities. This option is acceptable, but it also needs to 
>> modify VMI, and the modification is based on some Harmony specific 
>> contract, while my proposal above is based on public JNI spec, so 
>> this one is not preferred.
>>
>> Any ideas and comments are highly welcome.
>> [1]http://java.sun.com/j2se/1.4.2/docs/guide/jni/jni-14.html
>>
>>
>>
>
>


-- 
Paulex Yang
China Software Development Lab
IBM



---------------------------------------------------------------------
Terms of use : http://incubator.apache.org/harmony/mailing.html
To unsubscribe, e-mail: harmony-dev-unsubscribe@incubator.apache.org
For additional commands, e-mail: harmony-dev-help@incubator.apache.org


Mime
View raw message