Skip to content

CHANGELOG.md

Latest commit

 

History

History
223 lines (139 loc) · 8.6 KB

CHANGELOG.md

File metadata and controls

223 lines (139 loc) · 8.6 KB
Raw

Changelog

All notable changes to this project will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

Unreleased

Added

Removed

Changed

Fixed

Deprecated

Developer

v0.5.1

Changed

  • Improve http requests retry strategy. (115)
  • Optimizing performance of Dataset.download() (115)

v0.5.0

Added

  • Add Dataset.stac_catalog_size, and estimated_dataset_size properties. (#104)
  • Add descriptive __str__ methods to model classes. (#104)

Removed

  • Remove Python 3.7 support. (#106)
  • Remove Dataset.total_archive_size property, because of Dataset.download changes. (#104)

Changed

  • Rewrite Dataset.download() plus more robust downloading and filtering options. (#104)

Fixed

  • Remove outdated jupyter notebooks from the examples/ folder. (#104)

Developer

  • Remove tox from development environment. (#104)
  • Add Python 3.10 on Windows to CI test matrix. (#104)

v0.4.1

Fixed

  • Fix installation error by missing dependency: typing_extensions. (#79)

Developer

  • Check for missing dependencies in CI job. (#79)
  • Allow ReadTheDocs builds to succeed if warnings are emitted (#80)
  • Update trove classifiers to include Python 3.10 and bump dev status to "Beta" (#81)

v0.4.0

Added

  • client.list_models and client.get_model_by_id to discover ML Models by ID or by listing. (#71)
  • MLModel.list and MLModel.fetch methods analagous to new client methods described above. (#71)

Developer

  • Refactor of client and model python modules. (#71)
  • Configure Dependabot for automatic version upgrades of dev dependencies (#74)
  • Lock dev dependencies. (#75)
  • Remove Python 3.6 from tox config and add Python 3.10 (#73)

v0.3.1

Fixed

  • Use HEAD request instead of GET when initiating download (#68)

v0.3.0

Added

  • tags and text arguments to Dataset.list and client.list_datasets methods to allow filtering by datasets tags or free text search, respectively (#60)
  • client.get_dataset_by_id and client.get_dataset_by_doi methods to look up datasets by ID or DOI, respectively (#58)
  • Dataset.fetch_by_id and Dataset.fetch_by_doi methods analagous to new client methods described above (#58)
  • py.typed file included in package (#62)

Changed

  • client.get_dataset first attempts to get dataset using get_dataset_by_id then falls back to using get_dataset_by_doi if unsuccessful (#58)
  • Dataset.fetch uses client.get_dataset to first attempt getting dataset by ID, then falling back to fetching by DOI if unsuccessful (#58)
  • Upgraded PySTAC dependency to ~=1.1..0 (#63)

Removed

  • Support for Python 3.6 (#61)

Fixed

  • Numerous type annotation issues (#62)

Deprecated

Developer

  • Set different API root URL using MLHUB_ROOT_URL environment variable (#56)

[v0.2.2]

Fixed

  • Documentation example in Authentication docs (#53)

Developer

  • Moving to One Flow branching strategy instead of Git Flow for simplicity (#49)

v0.2.1

Fixed

  • api_key argument from Dataset.fetch used by Dataset instance for downstream requests (#48)
  • Dataset.download errors when api_key argument is passed (#48)

v0.2.0

Changed

  • Pins PySTAC to v0.5.4 (#43)

    Later versions automatically resolve links, which was leading to unnecessary network requests and issues with matching in VCR.py

Fixed

  • Allow user-defined profiles location (#27)
  • License file is now included in the distribution (#37)

Added

  • Properties to get archive sizes without downloading (Collection.archive_size and Dataset.total_archive_size) (#44)

  • New attributes on Dataset class (doi, citation, and registry_url) (#40)

  • Collection.registry_url property to get the URL for the Collection's registry page (#39)

  • Available as conda package via conda-forge (#34)

    $ conda install -c conda-forge radiant-mlhub

Developer

v0.1.3

Fixed

  • New attributes in the /dataset response will no longer break the Dataset class (#42)

v0.1.2

Fixed

  • Implicit dependency on typing_extensions (#29)

Developer

  • Manually caches properties instead of using functools.cached_property/backports.cached_property

v0.1.1

Added

  • Ability to resume archive downloads(#24)
  • Automatically retry requests that fail due to connection issues (#24)

v0.1.0

First working alpha release of the Radiant MLHub Python client.

Includes support for:

  • Configuring authentication using profiles or environment variables
  • Listing datasets and collections
  • Fetching datasets and collections by ID
  • Downloading collection archives