[code-reflection] RFR: Array view and PtrOp changes [v7]

[Gary Frost]

On Tue, 13 Jan 2026 20:30:53 GMT, Ruby Chen <duke at openjdk.org> wrote:

>> Implementing suggested array view and ptrOp changes, cleaning up code.
>
> Ruby Chen has updated the pull request incrementally with one additional commit since the last revision:
> 
>   Fixes

hat/core/src/main/java/hat/dialect/HATPtrOp.java line 54:

> 52:                             .map(fieldNode -> fieldNode.name)
> 53:                             .toList();
> 54:                     // remove the "array" field from the fields

So remove the last "array" field from fields   then remove my "is this .... " comment?

hat/core/src/main/java/hat/phases/HATArrayViewPhase.java line 61:

> 59:                 switch (op) {
> 60:                     case JavaOp.InvokeOp $ when invoke(lookup(), $) instanceof Invoke invoke -> {
> 61:                         if (invoke.namedIgnoreCase("add","sub","mul","div")) {

Oh good catch. Sorry about that.

hat/core/src/main/java/hat/phases/HATArrayViewPhase.java line 66:

> 64:                                     lookup(),
> 65:                                     invoke.name(),
> 66:                                     invoke.varOpFromFirstUseOrThrow().varName(), // varNameFromInvokeFirstUseOrThrow(invoke)

remove the comment?

hat/core/src/main/java/hat/phases/HATArrayViewPhase.java line 130:

> 128:                             Op.Result buffer = replaced.getOrDefault(r, r);
> 129:                             if (HATPhaseUtils.isVectorOp(lookup(),arrayLoadOp)) {
> 130:                                 Op vop = opFromFirstOperandOrThrow(buffer.op()); // resultFromFirstOperandOrNull(buffer.op())).op();

same remove all these comments where I suggested using helpers.

hat/core/src/main/java/hat/phases/HATArrayViewPhase.java line 250:

> 248:                         (res.op() instanceof JavaOp.ArrayAccessOp || res.op() instanceof JavaOp.ArrayLengthOp)) {
> 249:                         buffer = res;
> 250:                         indices.addFirst(res.op() instanceof JavaOp.ArrayAccessOp ? res.op().operands().get(1).result() : res.op().operands().get(0).result());

You might comment this.  I guess this is differentiated whether this is a store/load and then selecting operand?

take a look at OpHelper.resultOfOperandNOrThrow...

hat/core/src/main/java/hat/phases/HATArrayViewPhase.java line 253:

> 251: 
> 252:                 }
> 253:                 // I think we need to comment this.  Not so obvious.

You can remove my request for comment, if you feel the comment suffices ;)

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

PR Review Comment: https://git.openjdk.org/babylon/pull/814#discussion_r2690140029
PR Review Comment: https://git.openjdk.org/babylon/pull/814#discussion_r2690142515
PR Review Comment: https://git.openjdk.org/babylon/pull/814#discussion_r2690143727
PR Review Comment: https://git.openjdk.org/babylon/pull/814#discussion_r2690145091
PR Review Comment: https://git.openjdk.org/babylon/pull/814#discussion_r2690152859
PR Review Comment: https://git.openjdk.org/babylon/pull/814#discussion_r2690154400