Add Parameter for Enabling PCAN Auto Bus-Off Reset

[1atabey1]

Allows enabling of automatic hardware reinitialisation in case of a bus off event by adding the parameter auto_reset (defaults to False in order to not change current default behaviour)

[codecov]

Codecov Report

Merging #1345 (2ba1a10) into develop (c4f0789) will decrease coverage by 0.12%.
The diff coverage is 100.00%.

❗ Current head 2ba1a10 differs from pull request most recent head e508790. Consider uploading reports for the commit e508790 to get more accurate results

@@             Coverage Diff             @@
##           develop    #1345      +/-   ##
===========================================
- Coverage    66.18%   66.06%   -0.13%     
===========================================
  Files           86       86              
  Lines         9005     8964      -41     
===========================================
- Hits          5960     5922      -38     
+ Misses        3045     3042       -3     

[1atabey1]

Hmm looks like some tests are failing for the UdpMulticastBus and to be very honest I have no clue how that bus works and how I could have broken anything there... Can someone with a bit more experience have a brief look and help a newbie out? :)

(I noticed that @felixdivo has done a lot of work on the virtual UDP bus, do you have any idea what's going wrong here?)

[felixdivo]

Hm. No, I don't really have a clue. It wasn't an issue back when I developed and merged it. Perhaps, the CI platform changed. This has happened before with regards to changing IP(v6) support.

I'm not available to investigate this any further, at the moment at least. Sorry.

[1atabey1]

Okay no worries, I'll see if I can make some sense out of it!

[lumagi]

Hey @1atabey1, @felixdivo
I have the same failure reports on my pull request. I don't believe it has anything to do with the pull requests because for my second commit the test passed just fine. I rather think it is a race condition in the test runner that only fails badly part of the time.

I did some investigation, the failing tests all do the following:

1. Send out CAN message on bus 1 with arbitration ID i
2. Receive message on bus 2 and assert that it is equal to the original message
3. Clear out the reception queue of bus 1 by receiving one message non-blocking. This operation is supposed to clear out the datagram that is looped back to the sending socket during step 1
4. Increment arbitration ID of the original message to i + 1 and send it out on bus 2
5. Receive message on bus 1 and assert that it is equal to the original message with incremented arbitration ID i + 1

All tests fail because the arbitration ID of the message received during step 5 is still equal to i and not i + 1. The only reason I can think of is that the looped back datagram is not yet available in the reception buffer of bus 1 in step 3 and hence not cleared out. Then, during step 5 of the test, the code reads the message with arbitration ID i instead of i + 1 and fails.

Unfortunately, I am unable to reproduce this bug locally. I will try to create a new pull request to debug this further.

[zariiii9003]

Could you add the parameter to the __init__() method instead of using kwargs.get?
I do not know why some of the other parameters are not __init__-arguments but i thinks that's bad for readability.

[1atabey1]

sorry, was too busy last month unfortunately.. Thank you two for finishing this up @zariiii9003 @lumagi !