Severity score vs evilness

[hamishwillee]

1. Would it be better to replace evilness everywhere with "severity"?
2. How are severity scores calculated and how do they related to evilness?

Looking at the cut back output below I think that probably this was originally evilness and it has been renamed severity-score without removing the old markup. Correct?

I am assuming the severity scores at higher levels are just sums of the scores in the lower levels.

If it were me, I would:

• remove all mention of evilness
• at the lowest level, change from "severity-score" to "severity". The score should just be used for the sum.

Otherwise keep as is.

{
...
   "evilness" : 703,
   "severity-score" : 703,
   "tests" : {
      "Altitude Estimate Divergence" : {
         "description" : "This test will FAIL if various craft's Altitude estimates diverge",
         "evilness" : 70,
         "name" : "Altitude Estimate Divergence",
         "results" : [
            {
             ...
               "evidence" : [
                  "max-delta=4.250000 metres",
                  "delta-threshold=4.000000 metres",
                  "delta-time-threshold=0.500000 seconds"
               ],
               "evilness" : 10,
               ....
               "severity-score" : 10,
               "status" : "WARN",
               "timestamp_start" : 1436925978006204,
               "timestamp_stop" : 1436925979317506
            },
            {
               "duration" : 60.46092987060547,
               "duration-units" : "seconds",
               "evidence" : [
                  ....
               ],
               "evilness" : 10,
               "name" : "GPS_RAW_INT",
               "severity-score" : 10,
               "status" : "WARN",
               ...
            },
            {
....
....
         ],
         "severity-score" : 70,
         "status" : "WARN"
      },

[mrpollo]

hey @hamishwillee you are correct the attribute used to be called evilness, we moved to severity after some user testing, unfortunately we had some py apps already using dronekit-la at that point and we wished to keep backwards compatibility, this is something we should remove on a future release and give time to our developers to update.

[hamishwillee]

@mrpollo Thanks. In that case, can we at least break compatibility and made the fields "severity" for a particular test and "severity-score" for a cumulative value? Just makes more sense!

@peterbarker I appreciate that there are customers requiring compatibility here. How about we output "severity" as the bottom level definition as well as severity-score (and evilness). We can keep severity-score as the name for cumulative severity.

I would then document evilness and bottom level severity score as "deprecated" (and I will document evilness as such either way). For bonus points, providing a flag to only output the recommended output would be cool.

[peterbarker]

Having a separate numbers for the total and the contribution at a particular level makes sense. In hindsight, we could have had "severity-score" and "severity-score-total".

So, if we do want to see the contribution at each level.... perhaps "severity-score-here"? "severity-score-contribution"?

Just plain "severity" doesn't convey enough information IMHO.

[hamishwillee]

As long as we can differentiate individual and cumulative results in an obvious way I will be happy.

Just plain "severity" doesn't convey enough information IMHO.

I don't agree, but in any case, would be happy with "severity-score" and "severity-score-total".