Ambiguities that arose from the test cases

[lyxal]

Round 2 of everyone's favourite game. This time, the failed tests are to do with decisions regarding things like typing or edge cases that weren't previously thought of. That means that some things can be ignored/left as-is.

• ½: [""] -> ["", ""] (what to do with empty string?) (Improve the behaviour of ½ #1168)
• F: ["12345",["2","4","5"]] : "13" (should lhs be converted to list if rhs is already list?) (Improve the behaviour of F in regards to type handling #1169)
• O: [1213121, "1"] : 4 (should string be considered same as int?) (Make O work when given a string and number and vice-versa #1170)
• U: ["1",1] : ["1",1] (should "1" and 1 be considered equal in this context?) Someone forgot to format the test correctly.
• V: [234,"2",4] : 234 (should "2" and 2 be considered equal in this context?) (Make V work properly with strings and numbers #1171)
• ḭ: ["a",3] : "" (Why wouldn't it be ["a"]?) (Fix a test case for ḭ #1172) Looks like it's valid after all
• Ȧ: [[], [1, 2], 2] : [2, 2] (What even should happen here? What does assigning to an empty list do?) (Make assignment work on empty containers #1173)
• İ: [[1,2,3], [-1,0.1,1.9,"0"]] : [3,1,2,1] (That's a tricky one. What if there's a string in the index list?) (Improve the behaviour of İ #1298)
• Ṁ: [134,1,2] : 1234 (Should the result be a list or a number? What if it was "a" instead of 2?)
• Ẋ: [12,34] : [13,14,23,24] (Should those be numbers or lists?)
• Ẏ: [123,2] : 12 (Number or list?) (Make Ẏ and Ż leave numbers as numbers #1199)
• Ż: [123,3] : 23 (Same as above) (Make Ẏ and Ż leave numbers as numbers #1199)
• ⌊: ["-123"] : -123 (Should - be counted as part of the number?) (Make ⌊ account for decimals and negatives #1198)
• ∪: ["12",12] : ["1","2"] (How should strings and numbers interplay here?) (Make the tests for ∪ account for different types #1299)
• Ǒ: [1.125, 2] : -3 and [1.125, 3] : 2 (What rules for divisibility did you use? /srs is there a way of determining decimal divisibility?) I think integer order is fine here. (Change the tests for Ǒ to account for normal integer rules #1301)
• øḞ: ["234",2,4] : "234" (Depends on how strings + numbers should interact) current behaviour is fine. (Change some tests for øḞ #1312)
• øṄ: [234,"2",4,1] (Same as above) same as above (Change some tests for øṄ #1313)
• ÞS: [123] : [1,12,2,123,23,3] (Lists or numbers?) (Make ÞS respect original type #1297)
• Þ∨: [1234,"13"] : [2,4] (yet another query about type interaction) It needed some attention anyway (Improve the behaviour of Þ∨ #1314)
• Þ∩: (will be the same as above) (Improve the behaviour of Þ∩ #1316)
• Þ∪: (this one too) (Improve the behaviour of Þ∪ #1317)
• Þ⊍: (and this one) (Change the tests for Þ⊍ #1318)
• ÞK: [123456] : [123456, 23456, 3456, 456, 56, 6] (should those be numbers or lists?) (Make ÞK respect original type if numeric #1303)

[lyxal]

From Seggan:

[lyxal]

From emanresuA:

[Steffan153]

My opinions:

• ½: yes: two strings means ["", ""]
• F: yes: lhs is string, so keep a string
• O: yes
• V: no: should be 434 (as a number, not a string like it currently is)
• ḭ: no: should be "a"
• Ȧ: no: [0, 2, 2]
• İ: no: it should treat the first list as indices: [0.1, 1.9, "0"]
• Ṁ: yes: all inputs were numbers, so keep it a number
• Ẋ: no: make them lists
• Ẏ: yes: keep it a number
• Ż: yes: keep it a number
• ⌊: yes
• ∪: no: different types should be treated differently, so ['1', '2', 1, 2]
• Ǒ: no: those should return 0, I don't understand how -3 and 2 were calculated
  I'll do the digraphs later.

[cgccuser]

• ½: [""] -> ["", ""] is fine. For every other input, ½ outputs a list of length 2, so why not for "" too?
• F: ["12345",["2","4","5"]] : [1,3]. I think lhs should be converted to a list. If someone wanted the result to be a string, they can join rhs on "".
• O: [1213121, "1"] : 0. I don't think strings should be treated the same as numbers.
• V: "2" and 2 should not be considered the same.

• Ẋ: [12,34] : cartesian(range(12), range(34)). I think there are fewer situations where it would be useful to do the cartesian product of the digits and turn them back into numbers. On the other hand, I've needed cartesian products of ranges in the past.
• Ẏ: [123,2] : 12. Despite what I said for Ẋ, here, it'd be useless to do range(123)[:2] because you could just do range(2) instead. If you have numbers as digits set, it wouldn't be useless to get the list [1,2], but I think people would usually want 12 instead. If they really do want the list [1,2], that's fine because it would get likely get implicitly converted into that list while going through a later element anyway.
• Ż: [123,3] : 23. This should be consistent with Ẏ's behavior.
• ⌊: ["-123"] : -123. Yes, - should be counted as part of the integer. After all, it's "integer," not "natural number."
• ∪: ["12",12] : ["12",1,2] or ["12",1,2,3,...,12] depending on whether you've set the R flag (btw, I dislike the character ∪, it looks just like U)