Skip to content

Releases: ww-tech/primrose

v2.0.0

22 May 17:20
f4e69d0
Compare
Choose a tag to compare

This release is a Python 3.8+ compatible version of Primrose. There are also some bugfixes and documentation improvements.

  • Set py38 compatibility #120
  • Fix configuration error handling #118
  • Updated documentation #119

v1.0.14.dev38

23 Mar 22:13
Compare
Choose a tag to compare
v1.0.14.dev38 Pre-release
Pre-release

This pre-release is a Python 3.8-compatible version of Primrose, for future development purposes.

v1.0.14

28 Mar 20:10
3d45a57
Compare
Choose a tag to compare

This release includes some enhancements and bugfixes:

  • Adding sphinx install for docs build #109
  • Updating configuration.py node registration to register node from all collected files #110
  • Fixing logging statement in configuration.py #111
  • updating moto to fix dependency issue causing failed tests #114
  • update the notification docs to reflect recent changes in slack apps configuration #117
  • updating author and maintainer of package

v1.0.13

02 Jun 18:42
c101178
Compare
Choose a tag to compare

This release includes some enhancements and bugfixes:

Adding DoNothing node: #104
Cleanup of extraneous files
Adding message labels to error notifications: #102
Bugfix to avoid data=Nonetype errors: #101
Updating requirements versions
Environment variable substitution via jinja templating #94

v1.0.12

01 Sep 14:33
Compare
Choose a tag to compare

This release includes some enhancements to transformer pipeline nodes and a new transformer pipeline

  • feature - TransformerPipeline node to build transformer sequences from a config file
  • enhancement - The SklearnPreprocessingTrasnformer constructor can now accept a string instead of a preprocessor object which allows it to be used in a TransformerPipeline
  • enhancement - reformatted the whole repo using python black
  • enhancement - pre-commit hook added to enforce black formatting

v1.0.11

25 Aug 15:01
Compare
Choose a tag to compare

This release includes various bugfixes and slack client update:

Bugfix - fix networkx and matplotlib requirements due to interface breaking updates ( #83 )
Bugfix for notification error messages ( #82 )
Bugfix for specifying external module prefixes in config metadata ( #81 )
Feature - include ability to write arbitrary notifications from upstream nodes ( #80 )
Bugfix for categorical variables ( #79 )

v1.0.10

08 May 14:49
Compare
Choose a tag to compare
  • Updated pytest as well as the testing code to make better use of monkeypatching environments.
  • Updated error message handling to allow nested errors to be passed outside of the node that is being run
  • Updated R version and tests
  • Updated train test split default inputs to be more general

v1.0.9

13 Jan 17:07
Compare
Choose a tag to compare

Updates:

  • Create Slack notifications on error and as DAG nodes via the class primrose.notification_utils.SlackClient. Custom notification classes can be built by inheriting from the primrose.notifications.abstract_notification.AbstractNotification class. (#40)
  • New reader - primrose.readers.r_reader.RReader. Module with AbstractReader implementation able to read datasets from R using rpy2. (#34)

v1.0.8

18 Dec 19:50
3a7f0b3
Compare
Choose a tag to compare

Release notes:

  • Support for pickle serialization through new classes readers.deserializer.Deserializer and writers.serializer.Serializer. Planned deprecation of writers.dill_writer.DillWriter in a future release.
    • #21 Add pickle support
  • primrose version command added.
    • #22 Adding primrose version command
  • Automatic class path registry of external nodes using env variable PRIMROSE_EXT_NODE_PACKAGE. No longer necessary to register nodes manually via NodeFactory.
    • #23 Add class path config and dynamic import/registration of external nodes
  • Jinja templating for config file fragments. No longer substitute with $$FILE$$, instead opting for standard jinja syntax.
    • #25 Jinja templating config file fragments
  • Modular primrose installations. The base primrose is stripped of graphviz and postgres dependencies, making it more lightweight and easier to install. To install with the additional functionality, please use pip install primrose[postgres, plotting]
    • #27 Primrose Install Modification
  • Allow training_fraction configuration value in class pipelines.train_test_split.TrainTestSplit to be equal to 1
    • #39 Train test split updates