Consistently use `order` naming for `PyBuffer_*` functions

[nybblista]

The declaration of the PyBuffer_FromContiguous uses order as the name of its last parameter, which is fine for C and Fortran styles, but in its definition, it uses fort as the name of the parameter!.

I think we should rename fort to order and reflect that on the function documentation. Also, we should replace fort in the function comment.

cpython/Include/pybuffer.h

Lines 61 to 76 in 797c2c3

	PyAPI_FUNC(int) PyBuffer_FromContiguous(const Py_buffer *view, const void *buf,
	Py_ssize_t len, char order);
	/* Copy len bytes of data from the contiguous chunk of memory
	pointed to by buf into the buffer exported by obj. Return
	0 on success and return -1 and raise a PyBuffer_Error on
	error (i.e. the object does not have a buffer interface or
	it is not working).
	If fort is 'F', then if the object is multi-dimensional,
	then the data will be copied into the array in
	Fortran-style (first dimension varies the fastest). If
	fort is 'C', then the data will be copied into the array
	in C-style (last dimension varies the fastest). If fort
	is 'A', then it does not matter and the copy will be made
	in whatever way is more efficient. */

cpython/Objects/abstract.c

Lines 617 to 667 in 797c2c3

	PyBuffer_FromContiguous(const Py_buffer *view, const void *buf, Py_ssize_t len, char fort)
	{
	int k;
	void (*addone)(int, Py_ssize_t *, const Py_ssize_t *);
	Py_ssize_t *indices, elements;
	char *ptr;
	const char *src;
	if (len > view->len) {
	len = view->len;
	}
	if (PyBuffer_IsContiguous(view, fort)) {
	/* simplest copy is all that is needed */
	memcpy(view->buf, buf, len);
	return 0;
	}
	/* Otherwise a more elaborate scheme is needed */
	/* view->ndim <= 64 */
	indices = (Py_ssize_t *)PyMem_Malloc(sizeof(Py_ssize_t)*(view->ndim));
	if (indices == NULL) {
	PyErr_NoMemory();
	return -1;
	}
	for (k=0; k<view->ndim;k++) {
	indices[k] = 0;
	}
	if (fort == 'F') {
	addone = _Py_add_one_to_index_F;
	}
	else {
	addone = _Py_add_one_to_index_C;
	}
	src = buf;
	/* XXX : This is not going to be the fastest code in the world
	several optimizations are possible.
	*/
	elements = len / view->itemsize;
	while (elements--) {
	ptr = PyBuffer_GetPointer(view, indices);
	memcpy(ptr, src, view->itemsize);
	src += view->itemsize;
	addone(view->ndim, indices, view->shape);
	}
	PyMem_Free(indices);
	return 0;
	}

Linked PRs

• gh-137696: Rename 'fort' parameter to 'order' in PyBuffer_* functions #137866

[serhiy-storchaka]

This is an implementation detail that does not affect users in any way.

When someone rewrites this function for some other reasons, they can change the parameter name. Before that, it is just a code churn.

[nybblista]

@serhiy-storchaka
Yep, I agree with you that it will not affect users, but it still confuses the reader. Why do we use fort in the comment of the declaration to describe the order parameter? So, I think, in this case, fort as a name doesn't make sense, at least, in the context of the PyBuffer_FromContiguous declaration.

[serhiy-storchaka]

Well, this is a good reason to change this.

[CarlosEduR]

Could I work on it? I am looking for my first contributions to CPython...

[serhiy-storchaka]

If @nybblista does not already work on it.