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.

• Compatibility with Rails 7.2 and beyond

• Updated the max elasticsearch dependency version to < 8.7

• Revert the additional frozen_string_literal updates in 1.0.0 due to some issues with some classes. We'll update these in a separate release to isolate the problem.

• Consistently use quotations
• Update API to remove deprecation warnings for Elasticsearch 7.X.X in preparation for upgrade to 8.
• Freeze string literals
• Remove _type from query/create/update/delete methods
• Remove use_new_timestamp_format config
• Remove include_type_name_on_create config

• Don't publish pre-release builds to RubyGems

• Use refresh instead of flush where search is expected to be updated
• Drop support for Ruby 2.5 and 2.6
• Add support for Ruby 2.7 and 3.0, build against Ruby 3.1
• Set search_type to query_then_fetch as we no longer support Elasticsearch version < 7.0
• Set include_type_name when fetching the mapping. Will be removed in the next version to support ES version 8.
• Index names automatically use the new timestamp format when creating an index.

• expose 'matched_queries' property

• Compatibility with Rails 6

• Release on RubyGems using gem-publisher CircleCI Orb

• Release on Nexus using gem-publisher CircleCI Orb
• Packing gems
• bump the version for testing the Nexus repo

• remove superfluous 'to_ary' delegation that was causing issues downstream

• changes needed for use with Elasticsearch v7
• handle the v7 "total" as an object rather than a scalar
• use timestamp with no colons
• update elasticsearch gem version for consistency with target ES version
• expose refresh_index to force write (because in v7, flush no longer forces writes)
• allow for optional 'include_type_name_on_create' arg so that the :include_type_name can be passed

• use Arel.sql to avoid unsafe sql and eliminate deprecation warnings when used in Rails projects

• warn when configuring a index with subclasses if using ES version that does support them
• raise exception when creating or adding document to an index configured with subclasses if using an ES version that does not support them

• Give the option of retrying the deletion for certain exceptions during remap

• Fully clean up if error occurs during remap (assign all aliases back to original index)

• Adds support for preserving the order or normalized names of highlight through highlighted_attrs

• Adds support for passing arguments to Search definition through search(query, search_args) in index searching and msearch
• adds _explanation to hold value of returned explanations in the base document

• support action.destructive_requires_name setting by being explict about which indices to delete

• compatibilty with ES v6
• change mappings for 'index' to boolean. "string" type was replaced with "text"
• use "successful" from API response ('created' was removed)
• stringify keys for :mappings so clients can use symbol keys

• update remap to removing fields from the mapping that are not explicitly defined.

• fix search enumerator, missing first result set

• fix remap method to use the scan api properly.

• fix scan api to work with more recent versions of elasticsearch ruby.

• loosen support for elasticsearch-ruby versions to support more versions of elasticsearch

• Make Elasticity::Strategies::AliasIndex the default
• Use mapping instead of mappings, we wanna be consistent to ES not to elasticsearch-ruby
• Better automatic index name and document type

• add more response info to raised exceptions from reindexing

• update search and multi search interfaces to allow passing of general search definition arguments found in https://github.com/elastic/elasticsearch-ruby/blob/bdf5e145e5acc21726dddcd34492debbbddde568/elasticsearch-api/lib/elasticsearch/api/actions/search.rb#L125-L162

• update suggestions to pull from the proper key

• add next_page and previous_page to be compatible with will_paginate interface

• update multi search [] method to raise an exception with key name to make it easier to debug failed queries within a multi search hash.

• Change documents to be able to define indexes with multiple doc types. A Document class can define subclasses which are of different doc_types and all live in the same index's mappings.
• updated search queries to pass either a list of document types or a single document type.
• Update documents to generate a default document_type from the class name so that Documents always have a document type. You'll still usually want to manually define the document type, but it's no longer necessary.

• Add aggregations to multi_search

• Add ability to reindex individual attributes of a document using the bulk_update API.

• Refactor of multisearch and search facade
• Searches no longer return a simple array, they now return the Search::Results object which proxies array and is enumerable.

• add Search::Results proxy object so pagination and meta data methods can be standardize across all responses

• Fix issue with hash strings and pagination

• Added support for surfacing document _score on query results.