Device.read default datatype now np.float64 (was np.int32)?

[cweickhmann]

Is it possible that the default data type for a read command changed?

I have the SIS8300 example application with lots of device.read("XXX") not returning np.float64 where until 3.06 it was np.int32.
I could not directly confirm this from the changes in the source code.

[mhier]

I just checked the definition of Device.read() on 3.6:

DeviceAccess-PythonBindings/deviceaccess.py

Line 1415 in d81b2d4

def read(self, registerPath: str, dtype: np.dtype = np.float64, numberOfWords: int = 0,

It already defines np.float64 as default value for the dtype parameter.

The difference is that in the past this parameter was completely ignored. Now it is properly taken into account. On the other hand, the old implementation was actually smarter since it derived the returned data type from the register properties.

Maybe we should set the default value for dtype to e.g. None and change the implementation such that it will return the data type according to the register properties in that case. We could do that also for the register accessors.

@vaeng what do you think?