In 2D array getting an inner array by variable makes it impossible to get an elem of inner array

[Vulwsztyn]

Minimal reproducible example:

 (
  $arr := [[], [4, 5]];
  $i := 1; 
  $j := 1; 
  $arr[$i][$j];
)

It should be 1, but I get no match in excerciser.

$arr[1][1]; and $arr[1][$j] work.

Exploring other possibilities:

(
  $arr := [[1, 2, 3], [4, 5, 6], [7, 8, 9]];
  $i := 1;
  $j := 2; 

  $x := $arr[$i];
  {
      "double_variable":$arr[$i][$j],
      "one_variable": $arr[$i],
      "first_variable": $arr[$i][2],
      "second_variable": $arr[1][$j],
      "is_equal": $x[$i] = $arr[$i][$j],
      "separated": $x[$i]
  }
)

results in:

{
  "one_variable": [
    4,
    5,
    6
  ],
  "second_variable": 6,
  "is_equal": false,
  "separated": 5
}

[Turings98apprentice]

Edit:
@Vulwsztyn, it looks like you're just missing a colon...

Assignment operator is :=.
Equality operator is =.
$i is never initialized because it's compared to zero instead of being set to zero.

(
  $arr := [1];
  $i = 0; /* <<< should be $i := 0; */
  $arr[$i];
)

I missed this as well. Including my original (and incorrect) comment for posterity. Easy mistake to make!

It looks like what's happening is $i is being interpreted as a condition to filter the array rather than an index to specify one or more of its elements. Here's an even smaller minimal reproduction example:

(
  $arr := [1];
  $i = 0;
  $arr[$i];
)

... results in null / "no match" whereas ...

(
  $arr := [1];
  $arr[0];
)

... results in 1.

According to the documentation:

If the square brackets contains a number, or an expression that evaluates to a number, then the number represents the index of the value to select.

So this does appear to be a bug, since $i resolves to the number 0. I've tried it with other numbers as well, so it's not specific to the number 0 in particular.

Good catch!

[Vulwsztyn]

Thank you @Turings98apprentice I've adjusted the mre

[Turings98apprentice]

It looks like there's still a bug here, and the new coalesce operator is affected by it as well. I suspect this was a regression introduced in v1.2.6, because this bug doesn't exist in v1.1.1 or v1.0.10, but does exist in every version since.

Here's some additions to the minimal reproducible example that might help shed light on what's going on under the hood:

(
  $arr := [[], [4, 5]];
  $i := 1; 
  $j := 1;
  {
    '1 (Fail): $arr[$i][$j]': $arr[$i][$j],
    '2 (Pass): ($arr[$i])[$j]': ($arr[$i])[$j],
    '3 (Fail): $arr[$i][0][$j]': $arr[$i][0][$j],
    '4 (Pass): $arr[1][$j]': $arr[1][$j],
    '5 (Fail): $arr[$i][1]': $arr[$i][1],
    '6 (Pass): $arr[1]': $arr[1],
    "7 (Fail): $arr[1] ?? 'fallback'": $arr[1] ?? 'fallback',
    '8 (Pass): $arr[$i]': $arr[$i],
    "9 (Fail): $arr[$i] ?? 'fallback'": $arr[$i] ?? 'fallback',
    '10 (Pass): $arr[1][0]': $arr[1][0],
    '11 (Fail): $arr[$i][0]': $arr[$i][0],
    '12 (Pass): $type($arr[1])': $type($arr[1]),
    '13 (Pass): $type($arr[$i])': $type($arr[$i]),
    '14 (Pass): $type($arr[1][0])': $type($arr[1][0]),
    '15 (Fail): $type($arr[$i][0])': $type($arr[$i][0])
  }
)

... yields:

{
  "2 (Pass): ($arr[$i])[$j]": 5,
  "3 (Fail): $arr[$i][0][$j]": 5,
  "4 (Pass): $arr[1][$j]": 5,
  "6 (Pass): $arr[1]": [
    4,
    5
  ],
  "7 (Fail): $arr[1] ?? 'fallback'": 5,
  "8 (Pass): $arr[$i]": [
    4,
    5
  ],
  "9 (Fail): $arr[$i] ?? 'fallback'": "fallback",
  "10 (Pass): $arr[1][0]": 4,
  "11 (Fail): $arr[$i][0]": [
    4,
    5
  ],
  "12 (Pass): $type($arr[1])": "array",
  "13 (Pass): $type($arr[$i])": "array",
  "14 (Pass): $type($arr[1][0])": "number",
  "15 (Fail): $type($arr[$i][0])": "array"
}