Docker image/tag: semitechnologies/weaviate:0.22.8
See also: example docker compose files in English, German, Dutch, Italian and Czech.

WARNING: ACCIDENTAL BREAKING CHANGE IN THIS RELEASE

This release accidentally introduced a regression that somehow made it past the test suite. See #1173 for details. Do not use this release, but instead use the next one (0.22.9) where the regression is fixed!

Deprecations

• meta?=true/false in REST deprecated
  Instead use the new underscore props, e.g. ?include=_classification for classification information or ?include=_vector to display the vector or ?include=_vector,_classification for both. Estimated removal in 0.23.0

Breaking Changes

none

New Features

• Underscore-Prop _classification in REST and GraphQL (#1155)
  Display meta information about a classification (if an object was subject to a classification). This information could previously be shown in REST only using the now-deprecated ?meta=true. Instead you can now explicitly request this information using ?include=_classification. Additionally this information is now also available in GraphQL (previously not possible) using the _classification prop alongside the schema-defined props

• Underscore-Prop _interpretation in REST and GraphQL (#1156)
  Display meta information about how an object was interpreted during vectorization, i.e. which words were usable, how they were weighed and additional meta information about each concept, such as the occurrence frequency in the underlying contextionary.

  Note: This feature requires a contextionary version of at least ...-v0.4.12 which is used in the linked docker-compose files above

  The feature is available as an optional ("underscore") prop in REST using ?include=_interpretation as well as a GraphQL _interpretation prop alongside the schema-defined props.

Fixes

none

WARNING: ACCIDENTAL BREAKING CHANGE IN THIS RELEASE

This release accidentally introduced a regression that somehow made it past the test suite. See #1173 for details. Do not use this release, but instead use the next one (0.22.9) where the regression is fixed!