cc @juliusv

No strong opinion on this yet, but once we go that direction, we could also consider a whole lot more variations:

• The same between a range vector and an instant vector (I don't see how the other way around would work).
• The same for arithmetic and set ops (for example, you could also have used an and operator, like probe_duration_seconds[10m] and probe_success == 1, or maybe in other cases you'll want to multiply or add an offset to all values before averaging them.

What would it change for instant vector? For instant vector this should already work with current promql.

I mean between a range vector and an instant vector: foo[5m] > bar

I think we can limit this to scalar for now.

@roidelapluie I would like to work on this proposal.
Could you provide some more expressions I can analyse with, especially combination of operators like the one you put in the description above.
As I have only come across operator combination in logical expressions, would that suffice here. e.g. avg_over_time(probe_duration_seconds[10m] > 0 < 10) be equivalent to avg_over_time(probe_duration_seconds[10m] > 0 ) and avg_over_time(probe_duration_seconds[10m] < 10)

@roidelapluie, as per my understanding, when applying a scalar on to a range vector, does this illustration below depict the correct behavior / results or the expected outcome is something else. Request your feedback here, please.

Input: prometheus_http_requests_total[2m] actual samples

• prometheus_http_requests_total{code="200", handler="/api/v1/query", instance="localhost:9090", job="prometheus"}

  5 @123456
  3 @123476
  

• prometheus_http_requests_total{code="200", handler="/metrics", instance="localhost:9090", job="prometheus"}

   5 @123456
   2 @123476
   3 @123496
   23 @123116
   13 @123136
  

expected outcome prometheus_http_requests_total[2m] > 4

• prometheus_http_requests_total{code="200", handler="/api/v1/query", instance="localhost:9090", job="prometheus"}

  5 @123456
  

• prometheus_http_requests_total{code="200", handler="/metrics", instance="localhost:9090", job="prometheus"}

   5 @123456
   23 @123116
   13 @123136