Here is a rough idea how it can work:
Comparator = Callable[[Any, Any], int]
@ccall
def qsort(arr: Sequence[Any],
num_elems: int,
elem_size: int,
comp: Comparator) -> None:
pass
The types have to be adjusted. And we want to use it like:
def sort_f16_idx_val_ascending(iv_ptr: CPtr, num_elements: i32) -> None:
iv = ctypes.cast(iv_ptr, ctypes.POINTER(gd_lab_4_idx_val))
qsort(iv, num_elements, sizeof(gd_lab_4_idx_val), f16_idx_val_cmp_ascending)
where the f16_idx_val_cmp_ascending function is the comparator. And we want this to call the qsort function in C.
I think we have most of these pieces in LPython/LFortran, we just need to write a test for this and ensure everything works and fix any bugs.
Reactions are currently unavailable
Here is a rough idea how it can work:
The types have to be adjusted. And we want to use it like:
where the f16_idx_val_cmp_ascending function is the comparator. And we want this to call the qsort function in C.
I think we have most of these pieces in LPython/LFortran, we just need to write a test for this and ensure everything works and fix any bugs.