v0.2.2

The main goal of this release is to remove a dependency on the deprecated XTract service for https downloading of datasets.

Full Changelog: 0.2.1...0.2.2

0.2.0 -- Minor release for compatibility with Globus SDK 3

Add compatibility with Globus SDK 3. Additionally, add back in the f.run() functionality so users can run their DLHub models with the Foundry interface. Add visibility of DOIs to dataset summaries.

What's Changed

• Adding doi to list by @kjschmidt913 in #171
• Fix init auth by @ascourtas in #176
• update reqs for Globus SDK 3, and version to 0.2.0 by @ascourtas in #177
• Dev by @ascourtas in #178

Full Changelog: 0.1.2...0.2.0

(Yanked) 0.1.2 -- Patch to expose `f.run()` functionality

Re-added the f.run() functionality so that Foundry users can run DLHub servables using the Foundry SDK. Requires Globus SDK v2 and DLHub SDK version 0.10.0

EDIT: After release we discovered a bug where instantiating Foundry without providing authorizers resulted in an error; this was fixed in release 0.2.0, and this release was yanked

0.1.1 (minor patch)

Add joblib to setup.py and requirements.txt

0.1.0 - Minor release for robust dataset functionality

Overview

• includes updates from 2021 summer students
• test coverage and GHA support
• Foundry SDK fixes
• improved example notebooks
• OOP refactoring
• README updates and improvements

Breaking Changes

Loading datasets has changed. Users will specify index=mdf when instantiating Foundry:
f = Foundry(index="mdf")
and will load data like this:
f.load('dataset_name_or_DOI')
data = f.load_data()["split_name"]

Previously, users did not specify an index and loaded data as f = f.load("dataset_name").

0.0.17

Patch to make Foundry compatible with Globus SDK>=3.0, as well as MDF. Currently incompatible with DLHub.

0.0.16

• Bumping version in setup.py - previous conflicted with PyPI.
• Updating funcX requirements

0.0.15

• Fixes issue with loading HDF5 files from the wrong path
• Updates funcX requirement to be in line with the version needed.
• Reduces testing on main deployment to PR only

Pre-release -- patch for specifying index

Fix to allow specification of index of dataset (e.g. mdf-test vs mdf-prod)

Pre-release -- full adoption of new metadata schema, better testing

• DOI functionality
• atom finding dataset compatibility
• improved exception handling
• improved and functional testing
• full adoption of new metadata schema across code, data formats, test datasets, and notebooks
• fixed some small bugs