User defined method handle

[John Rose]

On Feb 5, 2009, at 2:04 AM, Rémi Forax wrote:

> BTW, the Java code of bound method can be simplified.
> In my opinion, a bound method handle is a direct method handle plus an
> object at argument zero.

Yes.  That is true, because there is no (net) stack motion, as the MH  
which is the receiver of the MH.invoke call is replaced by the bound  
receiver of the target method, and the other arguments do not need to  
be moved at all (on the JVM interpreter stack, or probably in any  
register-based calling sequence).

I suppose we can call this pattern "receiver replacement"; it is  
probably a typical elegance of bound method pointers in all sorts of  
systems, such as (I'm guessing) C#.

I made MHs.bind into a standard method mainly for convenience (to  
simplify otherwise tricky use cases).  On the other hand, the more  
general MHs.insertArgument optimizes the (common) case of receiver  
replacement, so MHs.bind and MHs.insertArgument are interchangeable  
for their overlapping usages.  At least, it's supposed to work this  
way.  If the Java code (in jdk/meth.patch) does not enable this  
optimization, it's a bug.

The JVM code also has special paths to support receiver replacement  
efficiently; see the flag OptimizeMethodHandles in hotspot/meth.patch.

-- John