What do you think? afaik the usage of null atom in Erlang code is much less common.
It is less common, but that's on purpose. For instance, if I were to decode null into the more "standard" undefined then there are issues when trying to detect if a key doesn't exist vs a key that exists with a value of null.
After thinking on this more I've decided that I'm not going to change this behavior. While I do see the argument for mapping the two conventions, null serves as a useful "no value" marker while undefined remains for Erlang to say "key doesn't exist".