xbee.wait_read_frame() timeout issue

[kenanelgaouny]

Hello,
it seems that xbee.wait_read_frame() is not timing out when a timeout is specified
Here's what i have tried:

data = xbee.wait_read_frame(1)

Traceback (most recent call last):
File "<pyshell#32>", line 1, in
data = xbee2.wait_read_frame(1)
TypeError: wait_read_frame() takes exactly 1 argument (2 given)

ser = serial.Serial('COM5', 9600, timeout=1)
xbee=DigiMesh(ser)
data = xbee.wait_read_frame() ##stuck here until a packet is recieved

any help would be appreciated !

[jamesleesaunders]

Is this the same issue described in #49 ?

[kenanelgaouny]

Yes they are related, but Issue 49 talks about the case where an incomplete frame is received. For this issue the timeout is for when no data is received at all

[jamesleesaunders]

One of the most common reasons for unpredicted frames is when the Xbee API settings are not correct.

What AP and AO settings have you got?

Try:
API Enable (AP): 2
API Output Mode (AO): 3

Depending on what the above is you may also need to play with the ‘escaped=true’ setting when when constructing in Python.

ser = serial.Serial('/dev/ttyUSB0', 9600)
xbee = ZigBee(ser, escaped=True)

Also see: #26

[jamesleesaunders]

Did you find the solution/fix?

[kenanelgaouny]

yeah it was more of a workaround, i used another older version of the library that can be found here: https://github.com/thom-nic/python-xbee, with this library the timeout works fine and the line
data = xbee.wait_read_frame(1)
doesnt not cause an error.

[jamesleesaunders]

Ah, interesting, we should probably work out what the difference is in the @thon-nic version and get the fix encorporated back upstream.

[XYR010]

Maybe something along the lines of this, i looked into the change history of xbee/thread/base.py and its predecessor: xbee/base.py, in neither was the version of _wait_for_frame with self and timeout=None as args, so probably it wasn't in there since the beginning

diff --git a/xbee/thread/base.py b/xbee/thread/base.py
index 3b60c8b..b47ec45 100644
--- a/xbee/thread/base.py
+++ b/xbee/thread/base.py
@@ -105,7 +105,7 @@ class XBeeBase(_XBeeBase):
         frame = self._wait_for_frame()
         return self._split_response(frame.data)
 
-    def _wait_for_frame(self):
+    def _wait_for_frame(self, timeout=None):
         """
         _wait_for_frame: None -> binary data
 
@@ -119,6 +119,10 @@ class XBeeBase(_XBeeBase):
         """
         frame = APIFrame(escaped=self._escaped)
 
+        deadline = 0
+        if timeout is not None and timeout > 0:
+            deadline = time.time() + timeout
+        
         while True:
                 if self._callback and not self._thread_continue:
                     raise ThreadQuitException
@@ -130,6 +134,8 @@ class XBeeBase(_XBeeBase):
                 byte = self.serial.read()
 
                 if byte != APIFrame.START_BYTE:
+                    if deadline and time.time() > deadline:
+                        raise TimeoutException
                     continue
 
                 # Save all following bytes, if they are not empty

[bp1222]

So after reading the PR & thon-nic addition it sure seems as though this is a feature request. From what I can tell, this implementation was designed to be used in one of two ways. 1) as a blocking call, where calls to wait_for_frame would block and return a frame when available. and 2) as an async callback, where the wait_for_frame would run in it's own thread, hitting some application defined callback when data is available. Assuming you'd be putting that into some threadsafe queue that could then be processed in your applications main loop.

That said, I don't necessarily have an issue with this fix. My only complaint would be that the threading mode, and tornado implementations would have different function signatures. We'd probably want to update the tornado side of this as well to properly raise the exception if a frame is not received within the timeout window.

[jamesleesaunders]

@bp1222 you make an excellent point about the signatures! Do you (or @XYR010) fancy making this change to the tornado implementation also?

[bp1222]

@jamesleesaunders I can next week sometime, unless @XYR010 if you're familiar/interested in applying this feature to the tornado side.

[XYR010]

I didn't work on the tornado version yet, but i will try. when reading through thom-nics version i noticed that he uses a new timeout once the start-byte was read, so that incomplete frames timeout, maybe something like this should be added as well later on 2018-03-15 21:48 GMT+01:00 David Walker <notifications@github.com>:

…

@jamesleesaunders <https://github.com/jamesleesaunders> I can next week sometime, unless @XYR010 <https://github.com/xyr010> if you're familiar/interested in applying this feature to the tornado side. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#59 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AGdE9nhqLvzHkcMUjebMXSIQWgzUMDz5ks5tetOEgaJpZM4QIlj3> .

[bp1222]

This issue has been merged into master.