Confusing definitions of errorQuery and totalQuery

[susenj]

The definitions of errorQuery and totalQuery according to the comments in the code explicitly mention:

// ErrorQuery is a Prometheus query that will get the number/count of events
	// that we consider that are bad for the SLO (e.g "http 5xx", "latency > 250ms"...).

I have a query that gives some timeseries data which I consider are falling in the bad category of events. Since, the definition says it's the count of those events, should I keep errorQuery as errorQuery: count(myQuery) or shall I keep it as errorQuery: myQuery itself? Which format is correct?

The same goes for totalQuery where I am not sure if I need to give an explicit count() or not? Could someone please clarify?

Some examples in the repo don't use count() and that's where the definition I quote above differs.