Changes
=======

0.10 - 15 Nov 2005
   - Only run POD tests for disttest
   - Require minimum version 0.3.0 for libsenna, due to some API changes
   - t/01-sanity.t failed due to the fact that new libsenna made some 
     enhancements in the checks in the underlying mechanism
   - t/02-morph.t failed due to the fact that new libsenna does a more
     wide search in sen_index_sel()

0.09 - 09 Nov 2005
   - Add POD tests
   - Add new docs

0.08 - 05 Aug 2005
   - Further tweak on Build.PL / Makefile.PL.

0.07_02 - 03 Aug @005
   - Fix del(): mixup in the order of execution. 
   - I don't know what this is, but M::B chokes on second attempt to
     build the XS code if lib/Senna.c exists. work around it.

0.07_01 - 02 Aug 2005
   - Fix rt #133933. This was caused by a hoopla between the senna
     developer and myself wrt the understanding of sen_index_info().
     This release works, but I wish to change it back to something saner,
     if libsenna can be fixed.
   - Change object from hash based to scalar based
   - Remove redundunt calls from Perl to XS
   - Add more error checks. Things will croak more now.
   - Add Build.PL, because I want Senna.pm to display a proper
     license when uploaded to CPAN.

0.07 - 29 Jun 2005
   - Fix calling get_*_state_hv() so that it uses aTHX_ properly.
     Pointed out by Lyo Kato

0.06 - 23 Jun 2005
   - Require senna that has senna-cfg. This will make life much easier for
     people reporting errors and such. (As of this writing, it requires
     the latest senna release from subversion)
   - Change Makefile.PL accordingly
   - add Senna::Cursor->as_list()
   - Senna::Index and Senna::Cursor was leaking memory. Fixed.

0.05 - 20 Jun 2005
   - Integer keys fix again. This time check at the cursor level, too
   - Fix the return value from replace()

0.04 - 20 Jun 2005
   - No code change. For some reason the test directory had been wiped
     from MANIFEST, and therefore it was not included in the distribution

0.03 - 20 Jun 2005
   - Integer keys were not handled properly (I misread the Senna source),
     and it was causing segfaults. This has been fixed.

0.02 - 06 Jun 2005
   - Doc tweaks.
   - Check return values.
   - This be 0.02.

0.02_05 - 06 Jun 2005
   - Add t/02-morph.t and t/03-ngram.t. Patch by Jun Kuriyama
   - Revise some return value codes.
   - Doc tweaks

0.02_04 - 04 Jun 2005
   - Fix segfaults when Senna::Cursor has no results.
     Reported by Jun Kuriyama

0.02_03 - 31 May 2005
   - *** Incompatible Change ***
     Senna::Index->create now takes key_size argument
   - Add SEN_VARCHAR_KEY, SEN_INT_KEY and SEN_ENC_* constants
   - Add encoding(), key_size(), initial_n_segments(), and flags()
     (note that you will need a very recent version of senna)

0.02_02 - 30 May 2005
   - Start mucking with new API
   - Add Senna::Index ->remove
   - Change semantics of Senna::Cursor 
   - Add Senna::Cursor->hits

0.02_01 - 11 Apr 2005
   - Apply patch from Hiroyuki Oyama (work with latest senna API)
   - No longer use Senna::Index->new(). Use open() and create() instead.
   - Export flags from Senna.pm
   - TODO: Design API around new Senna API
   - TODO: More tests, examples

0.01
  - Initial version