i2c.py, ValueError, out of range 0-256 in writing_bytes

[javl]

Not sure what causes this problem, but I'd though it might be useful to at least post an issue report. I tried using the quick2wire library with a cheap accelerometer by running the following code:

from i2clibraries import i2c_adxl345
adxl345 = i2c_adxl345.i2c_adxl345(0)

But immediately I get the following error:

Traceback (most recent call last):
  File "./test1.py", line 6, in <module>
    adxl345 = i2c_adxl345.i2c_adxl345(0)
  File "/home/pi/projectfolder/i2clibraries/i2c_adxl345.py", line 100, in __init__
    self.setActivityThreshold()
  File "/home/pi/projectfolder/i2clibraries/i2c_adxl345.py", line 178, in setActivityThreshold
    self.setOption(self.ActivityThreshold, intervals)
  File "/home/pi/projectfolder/i2clibraries/i2c_adxl345.py", line 230, in setOption
    self.bus.write_byte(register, options)
  File "/home/pi/projectfolder/i2clibraries/i2c.py", line 14, in write_byte
    writing_bytes(self.addr, *bytes))
  File "/home/pi/projectfolder/quick2wire-python-api/quick2wire/i2c.py", line 97, in writing_bytes
    return writing(addr, bytes)
  File "/home/pi/projectfolder/quick2wire-python-api/quick2wire/i2c.py", line 108, in writing
    buf = bytes(byte_seq)
ValueError: bytes must be in range(0, 256)

Printing these values, I noticed at least one of them indeed isn't in this range: -3

As a quick hack I changed the function in i2c.py so it clamps the values in byte_seq before returning the message, after which everything works perfectly.

def writing(addr, byte_seq):
    lst = list(byte_seq)
    for i in range(0, len(lst)):
        lst[i] = max(0, min(255, lst[i]))
    byte_seq = tuple(lst)
    buf = bytes(byte_seq)
    return _new_i2c_msg(addr, 0, create_string_buffer(buf, len(buf)))

[lurch]

IMHO your 'quick hack' is only hiding the bug, not fixing it ;-) Sounds to me like the real bug is that something in i2c_adxl345.py is passing in the value of -3 to an i2c function when it shouldn't be.

[javl]

That's why I called it a 'quick hack', not a fix ;) It might help others that run into this problem (I was using the latest NOOBS Raspbian raspberry pi image, like a lot of other people).

[malkowski]

Looks like the actual bug is in the way the i2c_adxl345 library handles negative accelerometer values during initialization. I bet if you flipped your accelerometer upside down and ran your code again, it would work ;) At least that's what I found.

I'd say the way the quick2wire API handles this error is great -- the stack trace points out the exact point of failure, once a bit of logging is added to figure out where the rogue value is coming from.

As to the problem you're seeing though, I just ran into the same issue and can offer a couple pointers that may help. I had to set the offset on my accelerometer, since it didn't actually report 1g when sitting on a flat surface. (See the datasheet for the correct register, or the source code of i2c_adxl345.py, though it's not implemented fully.) I also modified i2c_adxl345.py and changed lines from "if intervals < 256" to "if 0 < intervals < 256", which is hacky, but "if intervals < 256" is already a bit hacky...so... there you have it. Hopefully that helps, even if it's totally off topic. ;)