Release 0.51.0 of gordo

Server:

• Support reporting expected models (#911)

Builder:

• Add DatalakeIncompleteTransferException exit code (#937)
• Add batching to handle errors with log_batch calls (#934)
• Cleanup filter_rows.py (#908)
• Builder: Use un-cached runtime properties (#936)
• PostgresReporter: insert or update machines (#932)
• Update TimeSeriesDataset metadata (#912)
• Add assets to dataset resolvers (#903)
  • 1111-gfb
  • 1112-gfc
  • 1175-kri
  • 1760-hea

Release 0.50.0 of gordo

• Change total anomaly to mean of square (#929)

0.49.1

Bugfix release, fixing release pipeline in github actions.

0.49.0

Workflow:

• Add PVC to grafana, fixing persistence issue

Model builder:

• Update threshold calculation for DiffBasedAnomalyDetector (#924)
• Convert server-to-sql into a Reporter (#863)
• Add MlFlowReporter (#889)
• Raise more informative error on join_timeseries (#892)
• Feature gates for mlflow and postgres reporters (#891)
• Add 'dataset' key for build metadata (#902)
• Include features thresholds per fold in metadata (#919)

Server:

• Return 404 for metadata on non-existing model (#917)
• Return 404 req to download non-existing model (#918)
• Add server-version route (#888)

Generall:

• Rename pipeline_from/into_definition -> from/into_definition (#898)
• dataframe_from_dict: parse for int and sort (#900)
• Upgrade to pandas 1.0.0
• Fix scikit-learn future warnings (#923)

Release 0.48.3

• Update serializer & fix KerasRawModelRegressor (#874)
• Add high-priority to server deployment pod (#852)
• Optimize argo lint of generated workflows (#879)
• Remove gevent as explicit dependency (#881)
• Add InsufficientDataAfterRowFilteringError to row_filter / Change EXCEPTION_TO_EXITCODE (#880)
• Replace gevent with gthread (#886)
• Don't fail workflow if codecov fails (#887)

Release 0.48.2

New features ✨

• Add possibility to set n_splits (#864)
• Precheck workflow generate gives valid yaml (#851)

Cleanup 🛀

• Update outdated readthedocs config (#860)
• Make metadata/machine use more consistent (#854)

Bug fixes 🐝

• Update log_metadata tmp dir to local path (#862)

Dependency bumps 🤜 🤛

• Bump dependencies (#859)

0.48.1

📉 Grafana updates 📈

• Add aggregate-anomaly to Total Anomaly plot (#850)

🛠️ CI and general fixes ⚒️

• Add testformatting and allelse (#840)
• Add MasterCI specific workflow (#839)
• Fix MasterCI (#842)
• Fix gordo components naming in docs (#841)
• Remove unneccessary dependencies (#846)
• Use pytest tmpdir fixture (#849)

🆙 Dependency bumps ⬆️

• Bump pyyaml from 5.2 to 5.3 (#837)
• Bump azureml-contrib-run from 1.0.81 to 1.0.83 (#836)
• Bump numpy from 1.18.0 to 1.18.1 (#833)
• Bump numexpr from 2.7.0 to 2.7.1 (#835)

Release 0.48.0

• Rename gordo-components -> gordo (#831)

Requires renaming all imports using gordo_components to gordo.
Related commands such as gordo-components build ... are now gordo build ...

Release 0.47.0

Notable changes ⚠️

• Update tag-anomaly -> tag-anomaly-scaled references (#808)
• Add cv fold timestamps and n points to metadata (#785)
• Fix artifact logging with mlflow (#798)

Other changes and fixes 🔧

• Update CircleCI tests (#801)
• Use project revision instead of version (#804)
• Upgrade Python version to 3.7.5 (#591)
• Add .python-version @3.7.0 for Dependabot (#807)
• Add revision list view to server (#805)
• Add metadata fixture (#795)
• Endpoint to list models by revision (#809)
• Use try block in reading metadata (#823)
• Change GORDO_PROJECT_VERSION -> GORDO_PROJECT_REVISION (#827)
• Replace CircleCI with Github Actions (#821)
• Server: use revision to serve metadata/models (#810)
• Client: use revision from server (#816)
• Add dataclasses for machine.metadata (#819)
• Replace tests/utils globals with fixtures (#822)
• Include log_level as key in template (#762)

Dependency updates ⬆️

• Bump dependency versions (#825)
• Update dependencies (#814)

Release 0.46.0

Breaking change

• Make Machine a first class citizen (#794)

A major refactoring and renaming have taken place. Essentially the concept of Machine is more prominent.
For the users, the main difference is the paths to models.

For example:

from gordo_components.model.models import KerasAutoEncoder

is now

from gordo_components.machine.model.models import KerasAutoEncoder