Efficient Assignment

[GregYannes]

Assigning Once...

Because data may be nested arbitrarily deep, a recursive Function would return and assign the result arbitrarily many times. So instead we implement *Raw() as a Subroutine, to which we pass a variable v by reference:

Idx/src/Idx.bas

Lines 87 to 88 in f05dcb3

	' Index dynamically into the structure and extract the value there.
	IndexRaw x, v := Index, i := indices, l := low, u := up

Idx/src/Idx.bas

Lines 140 to 141 in f05dcb3

	' Index dynamically into the array and extract the value there.
	Arr_IndexRaw arr, v := Arr_Index, i := indices, l := low, u := up

Now *Raw() assigns the "return value" (to v) only once, when it reaches the "bottom".

Idx/src/Idx.bas

Lines 184 to 186 in f05dcb3

	' Base case: only one index left.
	Else
	Assign v, x(i(l))

Idx/src/Idx.bas

Line 202 in f05dcb3

Case 1: Assign v, a(i(l))

...Except Here

However, when it encounters an array, then IndexRaw() extracts its element into the placeholder v before recursing further into v itself.

Idx/src/Idx.bas

Lines 171 to 177 in f05dcb3

	' Extract the value from the array at those indices.
	Arr_IndexRaw x, v := v, i := i, l := l, u := k
	' Index (recursively) into that value using any further indices.
	If k < u Then
	IndexRaw v, v := v, i := i, l := k + 1, u := u
	End If

While this is visually elegant, is it introducing one further assignment per call? Should we instead implement Arr_Index() as a Function and rewrite this code like so?

	' Target the element from array at those indices: either extract its value as is...
	If Not k < u Then
		Assign v, Arr_IndexRaw(x, i := i, l := l, u := k)
		
	' ...or index (recursively) into it using further indices.
	Else
		IndexRaw Arr_IndexRaw(x, i := i, l := l, u := k), v := v, i := i, l := k + 1, u := u
		'        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
	End If