setOrbitsRepresentatives gives Nothing

[Jaxan]

Hi!

I am using the setOrbitsRepresentatives function now, but I run into a problem where it returns Nothing (and I am not sure it should do that...)
I am printing some debug info in my algorithm, and this is what I get (first line is the set, second line are the representatives given by setOrbitsRepresentatives):

{[a₁,a₂,a₃] : a₁ = a₂ for a₁,a₂,a₃ ∊ 𝔸, [a₁,a₂,a₃,a₄] : a₁ ≠ a₂ ∧ (a₁ = a₃ ∨ a₂ = a₃) for a₁,a₂,a₃,a₄ ∊ 𝔸, [a₁,a₂,a₃,a₄,a₅] : a₁ ≠ a₂ ∧ a₁ ≠ a₃ ∧ a₂ ≠ a₃ for a₁,a₂,a₃,a₄,a₅ ∊ 𝔸}
{Nothing, Just [0,0,0], Just [1,0,0,0], Just [1,1,0], Just [2,3,2,2], Just [2,3,2,3], Just [2,3,3,2], Just [3,4,3,5], Just [3,4,4,5], Just [3,4,5,3,3], Just [3,4,5,3,4], Just [3,4,5,3,5], Just [3,4,5,4,3], Just [3,4,5,4,4], Just [3,4,5,4,5], Just [3,4,5,5,3], Just [3,4,5,5,4], Just [3,4,5,5,5], Just [4,5,6,4,7], Just [4,5,6,5,7], Just [4,5,6,6,7], Just [4,5,6,7,4], Just [4,5,6,7,5], Just [4,5,6,7,6], Just [4,5,6,7,7], Just [5,6,7,8,9]}

However, when I try to reproduce the bug in ghci, the set is represented differently internally, so it doesn't show the bug...

Prelude NLambda> let x = NLambda.filter (\[a,b,c] -> a `eq` b) (replicateAtoms 3) `union` NLambda.filter (\[a,b,c,d] -> a `neq` b /\ (a `eq` c \/ b `eq` c)) (replicateAtoms 4) `union` NLambda.filter (\[a,b,c,d,e] -> a `neq` b /\ a `neq` c /\ b `neq` c) (replicateAtoms 5)
Prelude NLambda> x
{[a₁,a₁,a₂] : for a₁,a₂ ∊ 𝔸, [a₁,a₂,a₃,a₄] : a₁ ≠ a₂ ∧ (a₁ = a₃ ∨ a₂ = a₃) for a₁,a₂,a₃,a₄ ∊ 𝔸, [a₁,a₂,a₃,a₄,a₅] : a₁ ≠ a₂ ∧ a₁ ≠ a₃ ∧ a₂ ≠ a₃ for a₁,a₂,a₃,a₄,a₅ ∊ 𝔸}
Prelude NLambda> setOrbitsRepresentatives x
{Just [0,0,0], Just [1,0,0,0], Just [1,1,0], Just [2,3,2,2], Just [2,3,2,3], Just [2,3,3,2], Just [3,4,3,5], Just [3,4,4,5], Just [3,4,5,3,3], Just [3,4,5,3,4], Just [3,4,5,3,5], Just [3,4,5,4,3], Just [3,4,5,4,4], Just [3,4,5,4,5], Just [3,4,5,5,3], Just [3,4,5,5,4], Just [3,4,5,5,5], Just [4,5,6,4,7], Just [4,5,6,5,7], Just [4,5,6,6,7], Just [4,5,6,7,4], Just [4,5,6,7,5], Just [4,5,6,7,6], Just [4,5,6,7,7], Just [5,6,7,8,9]}

I will try to find a smaller example.

My current work around is to simply ignore the Nothing, but I'm not sure that is the right way to go.

[szynwelski]

Nothing in this case is a bug. Unfortunately I cannot reproduce it for both internal representations of the set. Could you send more information or all algorithm?