Suggested changes from posted PR:
- loosen the check for kNetworkNotFound for devices pre-1.7. Devices before 1.7 should tolerate an incorrect error (any error, but not success) and set a warning (use the record_warning feature to get this printed out at the end of the test). I think 1.7 we should enforce kNetworkNotFound. I think we'll get feedback / complaints during the test event and may need to adjust afterwards, but let's see how this goes as a first test.
- Eliminate the check for the malformed Network Key - as noted in the meeting, for thread the chances of having a bad network key on an otherwise matching network are slim. The common case would be a not found network. In field devices are definitely failing at this step and we should perhaps fix this in the SDK, but let's not let this block putting the test in
- fully commission the device at the end of the test so we don't have to wait out the failsafe.
- We should gate this test on the device reporting itself as having a concurrent connection (SupportsConcurrentConnection in the general commissioning cluster). I'd rather do this check within the test itself since we don't have a strong verification on the PICS, but I think I'll probably end up litigating this in the test event feedback.
Branch with some initial changes - https://github.com/cecille/connectedhomeip/tree/tc_cnet_4_24_proposal
Note that i just commented out the initial check on the xpanid rather than fixing it properly. Branch not properly tested, so use at your own risk, but it should give an illustration of the chnages and a starting point.
Suggested changes from posted PR:
Branch with some initial changes - https://github.com/cecille/connectedhomeip/tree/tc_cnet_4_24_proposal
Note that i just commented out the initial check on the xpanid rather than fixing it properly. Branch not properly tested, so use at your own risk, but it should give an illustration of the chnages and a starting point.