Unify & speed access to array-like objects behind ArrayXyzNode

[JaroslavTulach]

Currently the Enso engine is using InteropLibrary to read elements from array-like objects. However that's not enough - we have to guarantee that such elements are converted to Enso supported objects using HostValueToEnsoNode. As various incidents indicate:

• Splitting Gigantic EqualsNode into few Smaller Ones #6280 (comment)
• sort handles incomparable values #5998 (comment)

we often forget to do such conversion.

The standard Truffle way of avoiding such problem and also making the array-like object access fast is to define nodes that specialize according to the data (Vector, Array, foreign object). Let's define:

• ArrayLengthNode
• ArrayAtNode

and replace all array-like access with these nodes rather than using InteropLibrary directly.

[JaroslavTulach]

While investigating #7420 it turned out that eliminating need for HostValueToEnsoNode conversion can speed things up:

diff --git engine/runtime/src/main/java/org/enso/interpreter/runtime/data/Vector.java engine/runtime/src/main/java/org/enso/interpreter/runtime/data/Vector.java
index a966d47a5d..e9e283deaa 100644
--- engine/runtime/src/main/java/org/enso/interpreter/runtime/data/Vector.java
+++ engine/runtime/src/main/java/org/enso/interpreter/runtime/data/Vector.java
@@ -119,7 +119,7 @@ public final class Vector implements TruffleObject {
       }
       return WithWarnings.wrap(EnsoContext.get(interop), toEnso.execute(v), extracted);
     }
-    return toEnso.execute(v);
+    return v;
   }
 
   public static Vector fromArray(Object arr) {

the runtime/benchOnly caseBench3 benchmark is 10% faster. Eliminate need for this check for Enso only Vectors.

[JaroslavTulach]

The handling of warnings is also something that would better be unified as

• Avoid using a (warnings) Library on this #7525

demonstrates.

[enso-bot]

Jaroslav Tulach reports a new STANDUP for yesterday (2023-08-10):

Progress: - cleanup array-like classes: #7544

• CI on Mac is green: Fixing 'UTC' did not equal 'Etc/UTC' failure on Mac #7526
• TextBlocks improved: Text blocks in Enso grammar for TextMate based editor #7539
• benchmarking smallInteger.round: Implement Number.round as a builtin #7460 (comment)
• reviewed e-hern's PR: Allow configuring Project Manager server host and port #6774 (review)
• discussing [email protected]: https://graalvm.slack.com/archives/CNQSB2DHD/p1691669652147729?thread_ts=1691241276.560969&cid=CNQSB2DHD
• standup It should be finished by 2023-08-14.

Next Day: Array access unification

[enso-bot]

Jaroslav Tulach reports a new STANDUP for yesterday (2023-08-11):

Progress: - Array access unification - #7544

• review bench API: Ability to invoke all std benchmarks via jmh #7519 (review)
• emails & reviews & standups It should be finished by 2023-08-14.

Next Day: Finish array access unification

[enso-bot]

Jaroslav Tulach reports a new STANDUP for yesterday (2023-08-13):

Progress: - Array & Vector work: #7544 It should be finished by 2023-08-14.

Next Day: Finish array access unification

[enso-bot]

Jaroslav Tulach reports a new STANDUP for yesterday (2023-08-14):

Progress: - ArrayBuilder builtin: ae89884

• benchmarking
• standups & meetings It should be finished by 2023-08-14.

Next Day: Finish array access unification