V0.12.0

[ghost]

Client

• Introduced CogniteClient. Class with all api call methods made available through its instantiation. Ensures thread-safety. All sub clients (specific to APIs, e.g. files, events, timeseries, datapoints) inherit from APIClient class which give them access to sending requests through self.
• Expose http verbs with plumbing and error handling taken care of through CogniteClient to make it easier for users to hit endpoints which are not covered by the sdk.
• API Version is no longer exposed directly to user. 0.5 is default and some 0.6 features are exposed through an experimental sub client.
• Changed CogniteDataObject to CogniteResponse and added a string representation for pretty printing responses.
• Removed .to_ndarray() support for response objects
• Logging now defaults to log to DEBUG but log_level can be configured through CogniteClient object.

Restructuring

• All configuration is moved from config module to CogniteClient
• Everything requests related is moved to APIClient
• Preprocessing module removed
• Removed everything related to CLI which was abandoned
• _constants module removed
• config module removed

Documentation

• Improved overall documentation
• Added some simple examples for all queries.

Tests

• Tests no longer dependent on time series stored in mltest tenant

Data Transfer Service

• Updated to use new client interface

[codecov-io]

Codecov Report

Merging #175 into master will decrease coverage by 3.71%.
The diff coverage is 86.99%.

@@            Coverage Diff             @@
##           master     #175      +/-   ##
==========================================
- Coverage   91.38%   87.66%   -3.72%     
==========================================
  Files          29       19      -10     
  Lines        2344     1435     -909     
==========================================
- Hits         2142     1258     -884     
+ Misses        202      177      -25

Impacted Files	Coverage Δ
cognite/client/_utils.py	100% <100%> (ø)
cognite/client/stable/login.py	100% <100%> (ø)
cognite/client/experimental/analytics/__init__.py	100% <100%> (ø)
cognite/client/experimental/__init__.py	100% <100%> (ø)
cognite/client/exceptions.py	100% <100%> (ø)
cognite/client/experimental/analytics/models.py	50% <50%> (ø)
cognite/data_transfer_service.py	92.18% <50%> (+0.18%)	⬆️
cognite/client/experimental/time_series.py	65.62% <65.62%> (ø)
cognite/client/experimental/sequences.py	66.1% <70.45%> (ø)
cognite/client/stable/assets.py	85.13% <85.13%> (ø)
... and 26 more

[ghost]

Maybe since there have been such substantial changes to the interface, this should be a major, not a minor, version bump? @NilsBarlaug-cognite what do you think?

[NilsBarlaug-cognite]

Maybe since there have been such substantial changes to the interface, this should be a major, not a minor, version bump? @NilsBarlaug-cognite what do you think?

I guess we should normally do that, but since we're in major version zero it depends.

4. "Major version zero (0.y.z) is for initial development. Anything may change at any time. The public API should not be considered stable."
5. "Version 1.0.0 defines the public API. The way in which the version number is incremented after this release is dependent on this public API and how it changes."

(see https://semver.org)

So if we bump to version 1 that would mean we have officially released the first stable version and are not in initial development anymore. I would argue that it doesn't make sense to be major version 1 before the actual underlying API is stable (version 1), since we then can't guarantee that the SDK will not break. In addition, since there are so big changes now it will probably be followed by more changes and bug fixes the coming weeks. We should at least be somewhat stable before we go to major version 1.

What do you think @f1cognite?

[NilsBarlaug-cognite]

⭐️⭐️⭐️⭐️⭐️