Various clients all have to have numpy encoders for the json dumps, however they all seem to be copy/paste. This adds a generic one to .base and uses it for the l2client.
Is there any reason to not move all of the other encoders to the same common one? I don't see any custom logic in any of them right now.
Various clients all have to have numpy encoders for the json dumps, however they all seem to be copy/paste. This adds a generic one to .base and uses it for the l2client.
Is there any reason to not move all of the other encoders to the same common one? I don't see any custom logic in any of them right now.