Prerelease v0.1.2-alpha

Major changes

• Planner class - a convenience class creating and editing plans in Aquarium.
• Browser - a convenience class for browsing Samples and other models. Includes caching and query bundling to reduce
  number of queries and increase script speed.
• new queries (one, last, first)
• removed requirement of Marshmallow serialization/deserialization dependency, replaced by faster code
• Removed marshmallow as a serialization/deserialization library. A custom library is now used that avoids unnecessary
  field validation. As a result, model load and dump is many fold faster.

Minor changes

• Continuous validation - Cannot set Nested/Relationship attributes to arbitrary values. E.g. sample_type.sample = 5 will raise an Exception, but setting sample_type.sample = mysample will not.
  • sample_type.sample = mysample will also automatically set the sample_id attribute, since this is defined in the HasOne relationship as an attribute_key. But setting sample_type.sample.id = 5 will not change the sample_id key.
• setting None will no longer trigger a query attempt. In previous versions, a relationship whose value was set to None would intialize a query attempt anytime getattr accessed a model attribute, using round-about exception handling to avoid errors; this made it impossible to set a relationship attribute to None. Now, a query will only be initiated if the key does not exist in the models underlying data. This means that data received from Aquarium will be used as expected (i.e. setting {"sample": None} will not try to initiate a query next time model.sample is called).
  • FieldValue.set_value now handles None values. In previous version, None values passed into set_field_value would be ignored, making it difficult to reset a FieldValue. For example field_value.set_value(sample=None) will now reset the sample value for the FieldValue instead of being ignored.
• primary_key will return an id. If id==None, then the rid is returned (e.g. rid1023)
• HasOne relationship will automatically set the corresponding model reference when setting attributes. E.g. sample.sample_type = myst will also automatically set sample.sample_type_id = myst.id since this is defined in the HasOne field. Similar tracking is not implemented for other relationships.

Developer changes

• poetry https://poetry.eustace.io/ now used as the package manager, replacing pipenv
• vcrpy` is used to cache and store queries for deterministic testing.
• Improved documentation.