[Feature Request] Please implement a __setitem__ method for Tensor type

[cheburakshu]

Review Mojo's priorities

• I have read the roadmap and priorities and I believe this request falls within the priorities.

What is your request?

Please implement a __setitem__ method for Tensor that allows implicit indexing (just like how it has been implemented for __getitem__, see t[0, 0] in the example below) instead of having to use Index(m, n).

# Tensor multiplication

from tensor import Tensor
from utils.index import Index


var t = Tensor[DType.float32](2, 2)
t[Index(0, 0)] = 0
t[Index(0, 1)] = t[0, 0]
t[Index(1, 0)] = 2
t[Index(1, 1)] = 3

var t1 = Tensor[DType.float32](2, 2)
t1[0, 0] = 0

Output:

error: Expression [7]:27:7: no matching function in call to '__setitem__': 
    t1[0, 0] = 0
    ~~^~~~~~

Expression [0]:1:1: candidate not viable: callee expects 3 arguments, but 4 were specified
from memory.unsafe import Pointer
^

Expression [0]:1:1: candidate not viable: callee expects 3 arguments, but 4 were specified
from memory.unsafe import Pointer
^

Expression [0]:1:1: candidate not viable: callee expects 2 input parameters but 1 was provided
from memory.unsafe import Pointer
^

expression failed to parse (no further compiler diagnostics)

What is your motivation for this change?

Idiomatic usage of a Tensor/matrix.

Any other details?

No response