Environment

It seems that every version of jruby starting from at least 9k-pre1 has the following behavior.

Expected Behavior

2.3.1 :001 > [1, [1, []]].hash
 => 2397343324588010773
2.3.1 :002 > [2, [2, []]].hash
 => 4238421788679554352

Actual Behavior

jruby-9.1.0.0 :086 > [1, [1, []]].hash
 => 0
jruby-9.1.0.0 :087 > [2, [2, []]].hash
 => 0

We expect things to work in dry-validation because we use caching based on a hash of AST which in order is represented by a deeply nested array. The issue leads to returning the same result based on different outputs which surprises a lot.

Original values that you can use for testing a fix:

jruby-9.1.0.0 :001 > [[:and, [[:val, [:predicate, [:key?, [:age]]]], [:key, [:age, [:predicate, [:lt?, [23]]]]]]]].hash
 => 8481884717
jruby-9.1.0.0 :002 > [[:and, [[:val, [:predicate, [:key?, [:foo]]]], [:key, [:foo, [:predicate, [:lt?, [23]]]]]]]].hash
 => 8481884717

I don't know what exactly is going wrong but I tried to provide as minimal failing case as I could and it would supercool to have this fixed because you guys are doing a great job and having a support for jruby really matters for us, thank you!