2.0: Removed generic serialization of objects/iterables

[nzagorsky]

Hello! First of all thank you for such a great library!
After recent updates (most probably release of 2.0.x version) it started to fail with serialization error.

TypeError: �DY��� is not JSON serializable  

Steps to reproduce:

import ujson
ujson.dumps(set([1, 2, 3]))

or

ujson.dumps(set())

Thank you!

[hugovk]

Thanks for the report, that'll be this removal 53f85b1:

Removed generic serialization of objects/iterables

The behavior of ujson has always been to try to serialize all objects in
any way possible. This has been quite a deviation from other json
libraries, including Pythons standard json module, and the source of a
lot of confusion and bugs. Removing this quirk moves ultrajson closer to
the expected behavior.

Instead of trying to coerce serialization ultrajson will now throw a
TypeError: "repr(obj) is not JSON serializable" exception.

The "�DY���" bit is another issue: #258.

[bauman]

seems like this is bit controversial.

I'd like to express my support that this project should be limited to serializing and deserializing basic JSON objects as explicitly defined by RFC 7159 as fast and as resource efficient as possible. Generally accepted extensions like Inf and -Inf could stand to be left out as well.

ujson forks like nujson support extensions to numpy objects. Future forks can extend as needed into specific domains.

[murdav]

Personally, I liked to have the feature of generic serialization of objects/iterables. It was working great, especially when it is needed to serialize complex objects and save them into DB (JSONField in PostgreSQL).

[VMAtm]

This change is quite unexpected. We used this package for our REST API, our objects in memory are sets, and changing support for iterables pushes us to select another package.