ATM get's signature is

    def get(self, return_type='object', target=None, extensions=None,
            scope='all', regex_search=False, defined_fields=None, **kwargs):

which says that I cannot .get anything based on entities or metadata with fields which are get options, such as "target", "scope", etc. Even if ok now, any possible extension of get signature should be done really carefully (if at all) since might cause existing code using it break if such a new collision happen.

In the long(er) run I think it would be better to avoid such conflicts, so pybids API does not limit what BIDS terms etc could express. Possible solutions I see but may be there could be better ones

• define .gettter (the same signature as of get but without **kwargs) which would return the actual function to be called with **kwargs. get should then should just accept **kwargs only and implemented as a lean wrapper around return self.getter()(**kwargs)). May be for a while get could still accept any of the current options, complain about future deprecation, and pass them into the getter. Suggest better options for getter name ;)
  • a variant of this is to define a new (__call__) method which would provide the clean version of get(**kwargs) and slowly deprecate get itself
• extend get signature with an explicit query=None argument which would optionally take the value of **kwargs as a dictionary. May be while at it even could be replaced with a split metadata= entities= since now there is an internal battle (entities win) in case of a collision. Users then can just place their query arguments within a dict() to retain easy'ish way to specify, e.g. get(query=dict(suffix='T1w', return_type='file'))

But either of those two solutions would be incomplete if get with **kwargs would remain the documented/preferred way to query for files etc. So ideally get(..., **kwargs) way should get deprecated.

What do you think ?