For instance, the DocNotJSON error actually appears on a per-operation basis from KV. This might make sense since you can theoretically do a GET_DOC which doesn't parse the JSON. However, I'm not strictly sure that it makes sense to not just push the error up to the highest level if you do try to use the document like it's JSON when its not, since I can't imagine a use-case for doing a GET against a non-JSON document alongside a GET_DOC.
For instance, the DocNotJSON error actually appears on a per-operation basis from KV. This might make sense since you can theoretically do a GET_DOC which doesn't parse the JSON. However, I'm not strictly sure that it makes sense to not just push the error up to the highest level if you do try to use the document like it's JSON when its not, since I can't imagine a use-case for doing a GET against a non-JSON document alongside a GET_DOC.