Fixed minor errors

[d01c2]

This PR includes minor error fixes for updating to es2023:

• Fixed typo
• Error fix about numeric equality operation between mathematical value and ±Infinity

The result of this PR:

fail.json : 661 lines → 37 lines

1. Typo in rule.json in src/main/resources/manuals/default/:

   - "Return a new Symbol whose [[Description]] is _descString_.": "return (new descString)",
   + "Return a new Symbol whose [[Description]] is _descString_.": "return (new 'descString)",

   which occured many errors including "[InterpreterError] assertion failure: (? argument: \"Object\")". Found by checking INTRINSICS.Symbol.ir:

   def <BUILTIN>:INTRINSICS.Symbol(
   this: ESValue,
   ArgumentsList: List[ESValue],
   NewTarget: Object | Undefined,
   ): Unknown {
   if (< 0 ArgumentsList.length) let description = (pop < ArgumentsList) else let description = absent
   if (! (= NewTarget undefined)) {
   return comp[~throw~/~empty~]((new OrdinaryObject("Prototype" -> @EXECUTION_STACK[0].Realm.Intrinsics["%TypeError.prototype%"], "ErrorData" -> undefined)))
   } else {}
   if (= description undefined) {
   let descString = undefined
   } else {
   call %0 = clo<ToString>(description)
   let descString = [? %0]
   }
   return (new descString()) // should return (new 'descString)
   }

2. Some errors about equality operation between mathematical value and ±Infinity. For example:

   "[InterpreterError] wrong type for the operator ==: -1.0 and Infinity"

   This error could be fixed by adding some cases in eval function in src/main/scala/esmeta/interpreter/Interpreter.scala:

   + case (Equal, POS_INF, Math(_))     => Bool(false)
   + case (Equal, Math(_), POS_INF)     => Bool(false)
   + case (Equal, NEG_INF, Math(_))     => Bool(false)
   + case (Equal, Math(_), NEG_INF)     => Bool(false)

[stonechoe]

There are two remaining major errors:

• Unicode version problem: The latest unicode version is 15.0.0 but the current esmeta supports up to 14.0.0
• Mathematical value precision problem: "[InterpreterError] return not undefined: #4064"

1. Unicode version should be updated.
2. Test tests/test262/test/language/expressions/equals/bigint-and-number-extremes.js seems to be solved manually in src/main/resources/manuals/default/test262/categorized.json

[jhnaldo]

fail.json : 661 lines → 37 lines

Is it still a valid result?

I performed test262-test on this branch as follows:

esmeta test262-test -test262-test:progress -test262-test:concurrent -test262-test:log

And, the following summary is produced:

- time: 734,509 ms [12:14]
- total: 49,741
  - not-supported (N): 28,753
  - fail (F): 7
  - pass (P): 20,981
- pass-rate: P/P+F = 20,981/20,988 (99.97%)

Only 7 cases failed. Am I missing something?

[stonechoe]

I ran test262-test again with same command and retrieved the following summary with 27 fail tests (which is 35 lines total) of fail.json.

- time: 1,463,729 ms [24:23]
- total: 49,741
  - not-supported (N): 28,743
  - fail (F): 27
  - pass (P): 20,971
- pass-rate: P/P+F = 20,971/20,998 (99.87%)

By comparing fail.json with yours, the only difference was an issue of 'StackOverflowError'. I'm not sure... but maybe this result was occurred from the difference of testing environment.

[d01c2]

Now, we can retrieve following summary:

- time: 1,783,837 ms [29:43]
- total: 49,741
  - not-supported (N): 28,769
  - fail (F): 1
  - pass (P): 20,971
- pass-rate: P/P+F = 20,971/20,972 (100.00%)

with one remaining error according to fail.json:

{
  "No result when parsing failed" : [
    "language/destructuring/binding/syntax/property-list-followed-by-a-single-comma.js"
  ]
}

Should we resolve this error manually?

[jhnaldo]

I checked this test, and it failed when parsing the first parameter {x,} with ObjectBindingPattern. I will let you know how to debug parsing bugs later.

[jhnaldo]

I merged this PR. Thanks! 👍🏻