-
Type:
Bug
-
Resolution: Unresolved
-
Priority:
Major - P3
-
None
-
Affects Version/s: None
-
Component/s: None
-
Query Integration
-
ALL
-
None
-
None
-
None
-
None
-
None
-
None
-
None
Discovered during locust performance testing that this query:
$scoreFusion: {
input: {
pipelines: {
p1: {..., $score: {..., normalization: "minMaxScaler"}},
p2: {..., $score: {..., normalization: "minMaxScaler"}}
},
normalization: "none"
}
}
is much slower than:
$scoreFusion: {
input: {
pipelines: {
p1: {..., $score: {..., normalization: "none"}},
p2: {..., $score: {..., normalization: "none"}}
},
normalization: "minMaxScaler"
}
}
Which should be computationally equivalent and desugar to a similar pipeline.
We should understand why the first is so much less performant than the second, to both fix the issue, and understand if we have a flaw in our desugaring approach