We will add a unit test for this bug and then this should be merged.
Interestingly, we did a test in a real environment and didn't see any significant differences to the esp32-paxcounter. Due to more collisions I would have expected a higher difference between the max seen devices.
Note: the benchmarks we did are not affected as they used the bucket functions directly and did not go through mac_add. So the results should still be valid.
Well, that is embarrassing.
We will add a unit test for this bug and then this should be merged.
Interestingly, we did a test in a real environment and didn't see any significant differences to the esp32-paxcounter. Due to more collisions I would have expected a higher difference between the max seen devices.
Note: the benchmarks we did are not affected as they used the
bucketfunctions directly and did not go throughmac_add. So the results should still be valid.