Re: RFC: Require VM to provide JNI 1.2 support?

[Dalibor Topic]

Steven Augart wrote:
> Steven Augart wrote:
>
> > Mark has suggested that it might be a good idea for the next release
> > of GNU Classpath to require that the VM provide JNI 1.2.  He's asked
> > me to post this Request for Comments on the idea.
>
>
> Any objections to making it JNI 1.4 instead of JNI 1.2?  I believe it would
> not be much of a burden on VM Implementors; there are only three new 
> functions
> involved, and they can all be stubbed out.  On the other hand, there 
> appears to
> be the potential of major efficiencies.
>
> To quote from the Sun JNI spec:
>
> Three new functions introduced in SDK/JRE 1.4 allow JNI code to create, 
> examine, and manipulate direct buffers:
>
>     * NewDirectByteBuffer
>     * GetDirectBufferAddress
>     * GetDirectBufferCapacity
>
> Every implementation of the Java virtual machine must support these 
> functions, but not every implementation is required to support JNI 
> access to direct buffers. If a JVM does not support such access then the 
> NewDirectByteBuffer and GetDirectBufferAddress functions must always 
> return NULL, and the GetDirectBufferCapacity function must always return 
> -1. If a JVM does support such access then these three functions must be 
> implemented to return the appropriate values.

Sounds good for me.

cheers,
dalibor topic