v0.22.0rc2

• Fix some memory crashes (#1699)

Thanks @ManyTheFish for the fix! 🎉

v0.22.0rc1

• Fixes the String sort issue #1680
• Fixes the indexation issue #1677
• Improve error messages #1671 and #1685
• Change the position of the sort criterion (from 3th to 5th) #1678
• Fixes dumps related to the sort feature #1675

v0.22.0rc0

⚠️ Warning

• This release is a RC and should not be used in production
• We might have some issues for Windows users on this RC, this should be fixed in next RC

Breaking changes

• New custom ranking rules syntax. 
When updating the ranking rules:


"asc(title)"


becomes


"title:asc"


• New indexer: fixes some crashes and RAM consumption issues.
This feature needs a re-indexation of your documents. This will be done automatically when migrating your data from a previous version to this one. Check out how to update to the latest version of MeiliSearch

New changes

• Sort at the query time: set your sortableAttributes and then use the sort search parameter
  
⚠️ If you import your data via a dump, the sort ranking rules will probably not present. Check out this via the GET indexes/:ui/settings/ranking-rules route. If not, add the sort ranking rules to the third position using the POST indexes/:uid/settings route.
• Make the download-latest.sh script available for Windows user (using Cygwin or equivalent)

Misc

• Remove Sentry
• Remove Dependabot
• Improve Docker CIs
• Some updates in README

❤️ Thanks to the @core-team but also to: @singh08prashant, @CaroFG, @shekhirin and @okyanusoz

v0.21.1

Hotfix

• Compilation panic when no .git are present: #1638
  This issue was responsible for the Homebrew publication failure of the MeiliSearch binary

Thanks @Kerollmops for the fix!

v0.21.0

Breaking changes

• The index setting attributesForFaceting has been renamed to filterableAttributes

• The filters and facetsFilter search parameters are replaced by filter
  To use the filter search parameter, you must add the attribute you want to filter on to the filterableAttributes index setting.
  Only use the = syntax (e.g. genres = comedy). Filter expressions using the : syntax (e.g. genres: comedy) are deprecated. Filter expressions can be combined as strings or nested arrays:

  • string: 
"filter": "genre = comedy AND price > 13"

  • array: 
"filter": ["genre = comedy", "price > 13 "]


• When using the filter search parameter on an array, the operator != will only return results that do not contain the value

• The new default ranking rules are:


[
  "words",
  "typo",
  "proximity",
  "attribute",
  "exactness"
]

• fieldsDistribution has been renamed to fieldDistribution
• Rename MAX_MDB_SIZE environment variable and --max-mdb-size option flag to MAX_INDEX_SIZE and --max-index-size
• In the route /version, buildDate is now named commitDate and format has changed. MeiliSearch is now reproducible

Improvements

• New Web UI
• Facets with numbers
• Criterion words is available
• Search performance improvements
• attributesToHighlight now highlights matches inside arrays and objects
• Phrase search (" around words in a search query)
• Reduced RAM usage during indexation
• Removed the limit of 200 indexes per database
• Support typo on the first letter
• Facet distribution also works on arrays
• Enhanced logging capabilities with MEILI_LOG_LEVEL environment variable and --log-level option. Debug level now outputs query and search results
• Analytics (Amplitude and Sentry) are now unified and can be disabled at once with MEILI_NO_ANALYTICS and --no-analytics flag
• Improved exactness ranking rule
• A new update status, processing, has been added
• Add fields startedAt and finishedAt information on the route /dumps/{dump_id}/status.
• Cropping is now consistent
• Users can specify the payload limit size unit (e.g. Kb, Mb, Gb)
• Improve /stats route speed
• Improve compilation time with a caching strategy
• Renamed branch master into main.
• Improved README

Fixes

• Fix dumps bug that would exclude attributes not present in displayedAttributes from the dump file
• Fix error on snapshots when the user enters a wrong path
• Fix number of calculated hits when a distinct field is set
• Fix CORS error when authentication is missing
• Windows instances only take as much space as necessary now

v0.21.0rc7

• Fix decoding panic when deleting filterable attributes (#1601)

v0.21.0rc6

• Fix panic when indexing #1590

v0.21.0rc5

Fixes:

• Fix bug on facet values #1577
• Update README and issue templates
• Update dependencies (milli, tokenizer)
• Update telemetry link

Thanks @bb for reporting the bug!
Thanks @ManyTheFish for the fix!

💡 Visit our temporary docs for the v0.21.0. It's not completely finished yet, but it will help you try the v0.21.0 with your own dataset!

You can test this release by downloading the binaries available in this release.
Or you can use it with docker:

docker run -p 7700:7700 getmeili/meilisearch:v0.21.0rc5 ./meilisearch

v0.21.0rc4

Fixes:

• Dumps #1553
• Relevancy #1552

Thanks @CaroFG for reporting those bugs!
Thanks @ManyTheFish and @MarinPostma for the fixes!

💡 Visit our temporary docs for the v0.21.0. It's not completely finished yet, but it will help you try the v0.21.0 with your own dataset!

You can test this release by downloading the binaries available in this release.
Or you can use it with docker:

docker run -p 7700:7700 getmeili/meilisearch:v0.21.0rc4 ./meilisearch

v0.21.0rc3

Fixes

• Fix relevancy issue #1529
• Update the commit date format #1522
• Facets return the original value #1452
• Fix filterable attributes settings #1497
• Update limit of fields #1451
• Stop memory "leak" #1454
• FIx format error when sending documents #1535
• Add more tests

💡 Visit our temporary docs for the v0.21.0. It's not completely finished yet, but it will help you try the v0.21.0 with your own dataset!

You can test this release by downloading the binaries available in this release.
Or you can use it with docker:

docker run -p 7700:7700 getmeili/meilisearch:v0.21.0rc3 ./meilisearch