As stated in docs [0], JSON structures must not have duplicated keys.

>>> json.dumps({1:2, "1":3})
'{"1": 2, "1": 3}'

This is related to https://bugs.python.org/issue34972 . 



[0] "The RFC specifies that the names within a JSON object should be unique, ..." https://docs.python.org/3/library/json.html#repeated-names-within-an-object

Has this been a problem in practice, or just a theoretical issue?

To make this raise an exception, the JSON encoder would have to add one extra test per key.

I think we should just document the possibility.

JSON is JavaScript Object Notation, that is, a notation for JS Objects. Python dicts are much more general than that (not only in keys, but in values too: JSON keys must be strings, and values must be Strings, Numbers, Booleans, Arrays, (JS) Objects, or nulls -- both are restricted compared to Python dicts). There is really no reason to expect full embedding.

It's a theoretical issue, I didn't hit it myself.

The balance here is allow an invalid JSON to be created (but documenting that on some situations that will happen), or sticking to always produce valid JSONs, but with a hit in performance (which we don't know so far if it's big or small).

This module has been heavily used in the real world and we have no evidence that there is an actual problem to be solved.  Marking this as closed.  If a real world issue does arise, feel free to reopen and we can consider extending the API to have a validation step.