`BaseDocument.to_dict` serializes a document/embedded document into a dict,
which can be easily consumed by other modules (which in this case don't need
to be aware of MongoEngine-specific object types).
The output dict contains key-value pairs where:
* Keys are field names as they're defined on the document (as opposed to e.g.
how they're stored in MongoDB).
* Values are field values in their deserialized form (i.e. the same form that
you get when you access `doc.some_field_name`).
This more closely aligns with the rule that a single tests should test one
thing and one thing only. Previous code tested like 4 different things in a
single test and was hard to follow.
Previously, we were running the test suite for several combinations of MongoDB,
Python, and PyMongo:
- PyPy, MongoDB v2.6, PyMongo v3.x (which really means v3.6.1 at the moment)
- Python v2.7, MongoDB v2.6, PyMongo v3.x
- Python v3.5, MongoDB v2.6, PyMongo v3.x
- Python v3.6, MongoDB v2.6, PyMongo v3.x
- Python v2.7, MongoDB v3.0, PyMongo v3.5.0
- Python v3.6, MongoDB v3.0, PyMongo v3.5.0
- Python v3.5, MongoDB v3.2, PyMongo v3.x
- Python v3.6, MongoDB v3.2, PyMongo v3.x
- Python v3.6, MongoDB v3.4, PyMongo v3.x
- Python v3.6, MongoDB v3.6, PyMongo v3.x
There were a couple issues with this setup:
1. MongoDB v2.6 – v3.2 have reached their End of Life already (v2.6 almost 3
years ago!). See the "MongoDB Server" section on
https://www.mongodb.com/support-policy.
2. We were only testing two recent-ish PyMongo versions (v3.5.0 & v3.6.1).
We were not testing the oldest actively supported MongoDB/PyMongo/Python
setup.
This PR updates the test matrix so that these problems are solved. For the
sake of simplicity, it does not yet attempt to cover MongoDB v4.0:
- PyPy, MongoDB v3.4, PyMongo v3.x (aka v3.6.1 at the moment)
- Python v2.7, MongoDB v3.4, PyMongo v3.x
- Python v3.5, MongoDB v3.4, PyMongo v3.x
- Python v3.6, MongoDB v3.4, PyMongo v3.x
- Python v2.7, MongoDB v3.4, PyMongo v3.4
- Python v3.6, MongoDB v3.6, PyMongo v3.x
