Nov 26, 2013. Sphinx 2.2.1-beta and 2.1.3-release now available

Sphinx 2.2.1-beta, which incorporates several new major features, is now available. And, a maintenance release (Sphinx 2.1.3) has recently become available as well. Read on to learn more about both.

Sphinx 2.2.1-beta

The first release in the 2.2 series brings a number of new features, optimizations, removal of several legacy options as well as new default options. We’ve outlined the major feature additions below (and in this blog post), but for more information on what’s changed, consult the changelog.

Major features

  • English and German lemmatizers.
  • Integration with RLP for Chinese segmentation.
  • ALTER TABLE command for plain and RT indexes. Currently it’s only possible to add scalar attributes. More functionality will be added in the future.
  • New data source for indexing CSV/TSV files in the same fashion as for XML files.
  • New index type – template, which holds no data, but can be used for snippeting.
  • Improved DELETE command, which now allows arbitrary expressions in the WHERE clause (previously it was only possible to delete after ids)
  • N-best extension to GROUP BY clause
  • HAVING clause for SELECT statements which can be used for filtering on aggregate values
  • New ranking factors, bringing the total number to over 20!
  • Arbitrary JSON documents can now be stored in the JSON attribute
  • New advanced JSON searching functions ALL(),  ANY(),  INDEXOF()
  • GEODIST() now uses (by default) a lighter algorithm which is based on Polar Flat-Earth formula.  It’s now possible to specify the input/output unit type as well as the calculation algorithm to be used.
  • new ondisk_attrs, ondisk_attrs_default directives for choosing how attributes are loaded in memory


  • JSON attribute access is faster up to 2x times.
  • SELECTs are faster up to 3.5x times, depending on index schema size.
  • UPDATE is faster up to 3x times on big updates.
  • Internal optimization of threads table mutex, visible at 1000+ threads.

Removal and default new options

  • removed xmlpipe v1 data source (use xmlpipe2 instead), compat_sphinxql_magics directive, SetWeights() SphinxAPI call
  • dashes are not allowed anymore in index names
  • default SphinxAPI matching mode is SPH_MATCH_EXTENDED2
  • Sphinx is compiled by default with 64 bit IDs
  • default values for commonly used settings:
    • dict=keywords
    • charset_type=utf-8
    • enable_star=1
    • workers=threads
    • mem_limit=128M,
    • rt_mem_limit=128M

For a complete list please consult the  2.2.1-beta changelog.

Sphinx 2.2.1-beta can be downloaded from the beta download section.

Sphinx 2.1.3-release

Sphinx 2.1.3-release, a maintenance release for the 2.1 series, addresses a number of issues. For a complete list please consult the 2.1.3-release changelog.

Binary and source packages can be downloaded from the release download section.

Want more information?

Feel free to ask questions in the comments below. We will be happy to expand on any of these points.

Happy Sphinxing!

« »

3 Responses to “Sphinx 2.2.1-beta and 2.1.3-release now available”

  1. Home says:

    ver 2.1.3 sphinxQL update MVA always shows: “ERROR 1064 (42000): index ibb_sites: out of pool memory on MVA update”

  2. adrian says:

    @home: are you sure you don’t hit the mva pool limit? Otherwise fill a bug ticket.

  3. Home says:

    I don’t have many updates at all, pool limit=16MB (now changed to 32MB). Updates working but sometimes still gives error until restart searchd.

Leave a Reply