The `type<n>` cannot be hinted properly in IDEs.

[PyMarc2]

The type<n> cannot be hinted properly in IDEs.

e.g

import pyopencl as cl
import pyopencl.cltypes

N = 10
HOST_myArray = np.array([cl.cltypes.make_float4(1, 0, 0, 0)] * N, dtype=cl.cltypes.float4)

The make_float4 function and float4 types are not recognized. They are executed, but I think since they are not explicitly declared in cltypes.py, the ides have a hard time understanding

[inducer]

That's because they're generated at runtime:

pyopencl/pyopencl/cltypes.py

Lines 72 to 83 in 2916498

	names = ["s%d" % i for i in range(count)]
	while len(names) < padded_count:
	names.append("padding%d" % (len(names) - count))
	if len(titles) < len(names):
	titles.extend((len(names) - len(titles)) * [None])
	try:
	dtype = np.dtype(dict(
	names=names,
	formats=[base_type] * padded_count,
	titles=titles))

I suspect it'd be a simple exercise to rewrite this so that it generates source code that can then be read by an IDE? I'd be open to looking at a PR...

[PyMarc2]

I've looked everywhere and tried many things. It's impossible not to declare the variable name before runtime.
I would just add empty containers type<n>=None and make_type<n>=None.
It seems to work locally because after all the containers are overridden in the global namespace by the set_global from the create_vector_types function call.

Would that be an okay fix?

[inducer]

It's impossible not to declare the variable name before runtime.

I agree. That's why I proposed that we generate the source at install time.