RFR: 8023681: Fix raw type warning caused by Sink

[Brian Goetz]

Yes, this has been a beneficial side effect of lambda translation for 2+ 
years now.  The count of captured arguments is known statically to the 
metafactory and this (common) path can be special-cased:

             return new ConstantCallSite(MethodHandles.constant(samBase, 
inst));

The VM responds rewriting the call to the lambda factory as a constant 
load (and then further inlining and other optimizations can happen from 
there.)

On 8/23/2013 6:03 PM, Ali Ebrahimi wrote:
> nice. So we have currently this optimization?
>
>
> On Sat, Aug 24, 2013 at 2:14 AM, Brian Goetz <brian.goetz at oracle.com
> <mailto:brian.goetz at oracle.com>> wrote:
>
>     Because non-capturing lambdas are *already* constants!  The commonly
>     used trick, that was needed with inner classes to avoid
>     instantiating multiple instances of essentially identical objects:
>
>        static final Comparator<String> c = new Comparator() { ... }
>
>     is effectively implemented automatically (and lazily, instantiated
>     only on first use) by the language runtime for non-capturing
>     lambdas.  So the "extract into a static" trick can be relegated to
>     the scrap heap of "optimizations that don't optimize but still crap
>     up your code".
>
>     And, since the method castingIdentity() returns a constant, it will
>     be routinely inlined into a constant load.
>
>
>     On 8/23/2013 5:18 PM, Ali Ebrahimi wrote:
>
>         Why not to make castingIdentity method a constant?
>
>
>         On Fri, Aug 23, 2013 at 11:47 PM, Henry Jen
>         <henry.jen at oracle.com <mailto:henry.jen at oracle.com>> wrote:
>
>             Hi,
>
>             Please kindly review the fix for eliminate some warnings in
>             java.util.stream package.
>
>             Chained Sink is an internal implementation detail, add the
>             type for
>             downstream is more precise but verbose.
>
>             Included is also a couple other warnings cleanup.
>
>             http://cr.openjdk.java.net/~__henryjen/tl/8023681.0/webrev/
>             <http://cr.openjdk.java.net/~henryjen/tl/8023681.0/webrev/>
>
>             Cheers,
>             Henry
>
>
>