Some functions in Agg wrapper don't play nice with numpy dtypes #95
Description
I noticed this issue with GraphicsContextArray.clip_to_rect, but I imagine this is an issue in a few other agg functions.
This issue pops up when running one of the Chaco examples:
https://github.com/enthought/chaco/blob/master/examples/user_guide/grid_plot_container.py
The problem in the example is that the height is being set to an int, which gets converted to a numpy int at some point. A python int seems to cast correctly, but numpy ints don't:
import numpy as np
from kiva.agg import GraphicsContextArray
def python_ints(values):
return [int(a) for a in values]
def python_floats(values):
return [float(a) for a in values]
def test_clip_to_rect_dtypes():
gc = GraphicsContextArray((3, 3))
rect = np.zeros(4)
convert_funcs = [
python_ints,
python_floats,
np.float64,
np.float32,
np.int64,
np.int32,
np.uint8,
]
with gc:
for convert in convert_funcs:
try:
gc.clip_to_rect(*convert(rect))
except Exception as error:
print convert
print error
if __name__ == '__main__':
test_clip_to_rect_dtypes()