[code-reflection] RFR: Lower StringTemplateOp [v4]

[Paul Sandoz]

On Mon, 18 Mar 2024 10:32:02 GMT, Mourad Abbay <mabbay at openjdk.org> wrote:

>> The lowering of the StringTemplateOp. We chose not to have a block for every expression body, rather blocks will be created when necessary.
>
> Mourad Abbay has updated the pull request incrementally with one additional commit since the last revision:
> 
>   Lower each expression in a separate block

src/java.base/share/classes/java/lang/reflect/code/op/ExtendedOps.java line 2807:

> 2805:             also create a block for after expressions
> 2806:             *
> 2807:             * */

Suggestion:

            // Creates a block for each expression and a last block
            // to continue building from. An expression block branches
            // to the next expression block or the last block

src/java.base/share/classes/java/lang/reflect/code/op/ExtendedOps.java line 2826:

> 2824:                 if (exprIndex == 0) {
> 2825:                     b.op(branch(bb.successor()));
> 2826:                 }

Use a counted loop, rather than using `indexOf` inside the loop. You can hoist you the `b.op(branch(bb.successor()));` and place it in the above `else` branch of code.

src/java.base/share/classes/java/lang/reflect/code/op/ExtendedOps.java line 2884:

> 2882:         };
> 2883: 
> 2884:         private static Map<TypeElement, TypeElement> primitiveToWrapper() {

Recommend moving this functionality to `JavaType`. (You can store the result of Map.of in a static final field and reuse that rather than recompute on each call.) 

When we improve the modeling of Java types we can improve further.

src/java.base/share/classes/java/lang/reflect/code/op/ExtendedOps.java line 2888:

> 2886:                     BYTE, J_L_BYTE,
> 2887:                     SHORT, J_L_SHORT,
> 2888:                     JavaType.INT, J_L_INTEGER,

Suggestion:

                    INT, J_L_INTEGER,

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

PR Review Comment: https://git.openjdk.org/babylon/pull/39#discussion_r1528811192
PR Review Comment: https://git.openjdk.org/babylon/pull/39#discussion_r1528817633
PR Review Comment: https://git.openjdk.org/babylon/pull/39#discussion_r1528836728
PR Review Comment: https://git.openjdk.org/babylon/pull/39#discussion_r1528829873