Skip to content
This repository was archived by the owner on Jan 31, 2022. It is now read-only.

3.0
Compare
Choose a tag to compare
@clement-leprovost clement-leprovost released this 14 Apr 09:32
· 525 commits to master since this release
3.0
22990c1

This major version brings new features as well as bug fixes. In addition, a lot of refactoring has been performed
to achieve better maintainability and to align the Android client with the other Algolia API clients.

As a consequence, the public interface has changed in an incompatible way. Please refer to our
Migration guide for
detailed instructions.

New features

  • In-memory search cache: disabled by default, can be enabled per index.
  • Allow arbitrary query parameters to be specified: the Query class provides low-level, untyped accessors in addition
    to the higher-level, typed accessors.
  • Allow arbitrary HTTP headers to be specified (Client.setHeader)
  • Asynchronous requests are cancellable: asynchronous methods return a Request instance with a cancel method.
  • Low-level browse methods (Index.browse and Index.browseFrom)
  • New browse helper (BrowseIterator)
  • New method to clear an index (Index.clearIndexAsync)

Changes

  • Now only one type of asynchronous listener (CompletionHandler) with an interface containing only one method,
    used for both success and error cases. In addition to simplifying the code and aligning the Android client with other
    languages, this makes it possible to use Java 8 lambda expressions.
  • Remove operations requiring an admin API key. (The admin key should never be used on the client side.)
  • Align Query class parameters with the REST API
  • Improve the type safety of many Query parameters, especially the geo search-related parameters and the filters
  • Remove obsolete (and unused) enableDsn and dsnHost initialization parameters
  • Remove accessors to deprecated HTTP headers
  • Space character is now URL-encoded as %20 instead of +
  • Better naming convention:
    • Rename APIClient class to Client
    • Rename *ASync to *Async
    • Remove prefix on enum values

Fixes

  • Search queries now use POST instead of GET, working around any potential hard-coded limit to the URL string.
  • Delete by query uses browse instead of search, for both better performance and to work around any index settings
    (like distinct) that would silently prevent the query from selecting all objects.
  • More consistent error handling
  • Fix retry logic in case of host failure
  • Alleviate critical section on internal request methods

Misc. improvements

  • Use @NonNull annotations to document mandatory parameters and return values
  • Asynchronous methods no longer throw (declared) exceptions
  • Update documentation (a lot!)
  • Add test cases
Assets 2
Loading