InStream.get_channels_buffer_size uses wrong encoding

[bkeryan]

InStream.get_channels_buffer_size and OutStream.get_channels_buffer_size read the Task.channel_names property to calculate the maximum buffer size for a fault condition status property such as overcurrent_chans.

The problem is that these functions are calculating the length of the channel_names property after decoding from bytes to str. If Python represents str as UTF-16 or UTF-32, the number of characters in a str may be smaller than the number of UTF-8 bytes used to return that str from the DAQmx C API. When this happens, reading the fault condition property will likely return a "buffer too small" error when there is a fault.

Some potential solutions:

• Use DAQmxGetTaskChannels(task, NULL, 0) to query the size of the TaskChannels property in the C API's native encoding (UTF-8 or MBCS). This would also be a small optimization because it would query the size of the TaskChannels property without querying the actual value.
• Convert the task channels back to the C API's native encoding. This is a hack, so the first solution is better.

[bkeryan]

If Python represents str as UTF-16 or UTF-32, the number of characters in a str may be smaller than the number of UTF-8 bytes used to return that str from the DAQmx C API.

Correction: it doesn't matter what the internal representation of str is. What matters is that len() on a str counts Unicode code points, not UTF-8 bytes:

>>> x="é"
>>> x
'é'
>>> len(x)
1
>>> x.encode('utf-8')
b'\xc3\xa9'
>>> y="\u2603"
>>> y
'☃'
>>> len(y)
1
>>> y.encode('utf-8')
b'\xe2\x98\x83'
>>> z="\U00010437"
>>> z
'𐐷'
>>> len(z)
1
>>> z.encode('utf-8')
b'\xf0\x90\x90\xb7'