Added
- Support for all endpoints in Cognite API
- Generator with hidden cursor for all resource types
- Concurrent writes for all resources
- Distribution of "core" sdk which does not depend on pandas and numpy
- Typehints for all methods
- Support for posting an entire asset hierarchy, resolving ref_id/parent_ref_id automatically
- config attribute on CogniteClient to view current configuration.
Changed
- Renamed methods so they reflect what the method does instead of what http method is used
- Updated documentation with automatically tested examples
- Renamed `stable` namespace to `api`
- Rewrote logic for concurrent reads of datapoints
- Renamed CogniteClient parameter `num_of_workers` to `max_workers`
Removed
- `experimental` client in order to ensure sdk stability.