Review request for JDK-8027236

[Pavel A. Stepanoff]

SQE is OK with this fix

Pavel

On 01.11.2013 18:22, Attila Szegedi wrote:
> Please review JDK-8027236 at http://cr.openjdk.java.net/~attila/8027236/webrev.00
>
> This change makes sure that ConsString is no longer observable from outside the Nashorn engine internals.
>
> I also made some improvements to conversion mechanism where I spotted them. Namely, JSType.toJavaArray() used to always do getArray().asObjectArray() and then convert that to target type. This was very inefficient in cases where we had an IntArrayData/LongArrayData/NumberArrayData, and the requested array type was int[]/long[]/double[], respectively -- in these cases we could've just used a direct copy of the underlying array. Funnily enough, even though we had this mechanism in place in form of ArrayData.asArrayOfType(...), it was just sitting there unused; I guess we accidentally cut it out at some point.
>
> Most small changes to various classes have to do with the fact that whenever we used to go directly to the BeansLinker for some functionality (e.g. bound methods, static method linking etc.) we now must make sure that we route those calls through the new ConsString-converting logic.
>
> I'm still not 100% happy with the result, I feel like Dynalink should have a facility for linkers to declare such filters for internal objects that should never leave the engine and figure out where to apply them, but I don't think I have enough time in the Java 8 timeframe to figure out the correct design (but once I do, we can easily retrofit).
>
> Thanks,
>    Attila.