[code-reflection] RFR: Support conversions of primitives in testing context [v9]

[Paul Sandoz]

On Wed, 30 Oct 2024 00:28:42 GMT, Mourad Abbay <mabbay at openjdk.org> wrote:

>> Support conversions of primitives in testing context.
>
> Mourad Abbay has updated the pull request incrementally with one additional commit since the last revision:
> 
>   Better names for methods in TestPrimitiveTypePatterns

src/java.base/share/classes/java/lang/reflect/code/op/ExtendedOp.java line 3179:

> 3177:                                 || isWideningAndNarrowingPrimitiveConv(ps, pt)) {
> 3178:                             // e.g. int -> byte, narrowing
> 3179:                             // e,g. int -> float, widening that need check

Suggestion:

                            // e,g. int -> float, widening with need check

And rename `isWideningPrimitiveConvThatNeedCheck` to `isWideningPrimitiveConvWithCheck`.

src/java.base/share/classes/java/lang/reflect/code/op/ExtendedOp.java line 3187:

> 3185:                     }
> 3186:                 } else {
> 3187:                     // e.g. byte -> Byte, Number -> Double, ...

I think the primitive -> ref case, e.g., byte -> Byte, is not supported correctly. We need to box, which I believe is unconditional if from valid source. You can have another `else if (s instanceof PrimitiveType ps) { ... }`.

The boxing test is likely passing because the interpreter naturally boxes all primitive values and we don't differentiate between a primitive value and its box. Consider that a pragmatic limitation.

-------------

PR Review Comment: https://git.openjdk.org/babylon/pull/265#discussion_r1823435015
PR Review Comment: https://git.openjdk.org/babylon/pull/265#discussion_r1823488435