3.3.21

* bugfix: fixed select command to return correct from and to tags
* bugfix: fixed clear command to clear numeric lookup tables
* bugfix: fixed search for phrases with numbers and/or dates
* bugfix: fixed select command wildcard matching
* feature: cluster support for siets-load tool
* feature: numerous improvements in crash-safety during indexing
* bugfix: fixed background indexing slowing down command processing
* bugfix: fixed similar command

=============================================================

3.3.20

* feature: ability to turn off date and/or numer indexing and searhing
* bugfix: added command 'select' to API GET method
* bugfix: fixed command 'similar' realisation in API GET method
* feature: added ability to turn on/off snippet and highlight during search
* bugfix: small fixes in date parsing
* feature: ability to display begining of text, is snippet cannot be composed
* feature: ability to customize xml tag highlighting
* bugfix: fixed snippet and highlight if xml tags are specified close to plain text query

=============================================================

3.3.19

* bugfix: whitespace in numeric search
* bugfix: memory leaks in alert handling and command dispatcher
* bugfix: national search not searching all words
* bugfix: adding alert with frequent (ignored) word
* feature: alert handling improved
* feature: clear command clears alerts as well
* bugfix: error messages in alerting
* feature: document parsing timeout created
* bugfix: command dispatcher hanging fixed under heavy load
* bugfix: search with all negative terms

=============================================================

3.3.18

* feature: support for indexing of UCS graphical signs (internationalization issue)
* bugfix: document deletion after clear command
* bugfix: problems in alerting (phrases in combination with nested operations)

=============================================================

3.3.17

* bugfix: case sensitivity fixed in alerting
* bugfix: cluster result composition fixed (bug introduced in 3.3.15) 
* feature: in alerting - filter expression returned along with trigger id

=============================================================

3.3.16

* bugfix: growing repository (DOC) buckets
* bugfix: locking issues
* bugfix: search for a last word in document caused error in some cases

=============================================================

3.3.15

* bugfix: memory leak in recover command for deleted documents
* bugfix: word count overflow
* feature: .NET interface added to API

=============================================================

3.3.14

* bugfix: temporary bucket handling which could cause index corruptions due to restarts during indexing
* feature: delete and update commands optimized
* bugfix: stemming fixed (bug introduced in 3.3.12)

=============================================================

3.3.13

* bugfix: date/number parsing searched incorrectly in case of specsymbol overload
* bugfix: NOT operation and phreses in alerting fixed

=============================================================

3.3.12

* bugfix: date/number parsing caused crashes in rear cases
* feature: alerting (beta testing)

=============================================================

3.3.11

* feature: XML-drilldown functionality (beta testing)
* bugfix: fixed unremoved lock in timeout situation

=============================================================

3.3.10

* feature: date range searches available
* bugfix: performance issue concerning large data-sets in search-ids command (affecting cluster version as well)
* bugfix: siets-load produced fake error messages in log file
* bugfix: ascending order for interval filters in numeric search

=============================================================

3.3.9

* bugfix: bugs regarding indexing fixed (problem fixed in v3.3.7 intorduced another problem)
* bugfix: some bugs in cluster result merging fixed (in case if not all nodes has results and <more> tag value)

=============================================================

3.3.8

* bugfix: returning result above 1000 if max_resultset is configured

=============================================================

3.3.7

* bugfix: deleting large volumes of data caused crashes in some cases
* bugfix: result relevancy ordering fixed (was incorrect with custom weight tags and OR boolean expression in some cases)

=============================================================

3.3.6

* bugfix: locking revisited (issues on dual-processor servers)

=============================================================

3.3.5

* bugfix: critical bugs fixed in number parsing, date searches freezed

=============================================================

3.3.4 (not released due to critical bugs)

* feature: application tag as HTTP GET parameter
* bugfix: similar document search returned irrelevant results in some cases (since 3.2.8)
* bugfix: numeric and data indexing - produced invalid lookup tables (harmless but taking disk space)

=============================================================

3.3.3-2

* bugfix: search for month abreviation 'jan', 'feb', ... together with stemming

=============================================================

3.3.3

* removed dependecy from libunicode
* bugfix: few words were lost from index in case of restart of server without clean shutdown during indexing
* bugfix: in very rear cases docd failed on large (100 on page) result sets
* feature: date range searches added

=============================================================

3.3.2 

* API for numeric searches is fully developed and complient with documentation.
* Tag compression added. Useful for small or tag-intensive documents.
* Merged with bugfixes form 3.2.10

=============================================================

3.3.1 (beta)

Lots of bug fixes and optimisation.
Note: data format for range-searches not compatible to that
of previous version.

=============================================================

3.3.0 (alfa)

Branch from 3.2.9. Added functionality: numeric range searches.
This version is alfa test release.

=============================================================

-------------------------------------------------------------

=============================================================


3.2.10

* bugfix in process queue (maximum parallel process constrain accessing inverted index) under transaction overload

=============================================================

3.2.9 

* bug in cluster result merging fixed (resulting incorrect ordering searching with relevance and wihout domain grouping)
* bugs fixed in integrated file conversion
* LIST-LAST command added to list documents in physical order

=============================================================

3.2.8

* stability improvements (timeouts in mtxd, maximum parallel process limit and transaction queueing)
* word stemming fixed (for multi-word queries)
* memory leak in STATUS command fixed
* integrated file convertion from binary file formats to Siets XML document implemented in API

=============================================================

3.2.7

* config parameter override form API commands added
* config parameter added for minimum length of wildcar pattern prefix
* bug fixed in <realquery> tag: fixed to work properly also with meta-tagging

=============================================================

3.2.6

1) SELECT command was added to retrieve document IDs matching given wildcard pattern

2) following bugs were fixed:
* snipperts and highlighing with word stemming feature - all word forms now are highlighted
* whitespace in document IDs, before this documents that contained whitespace in IDs was lost after restart of server instance resulting in failures of RETRIEVE and DELETE of these documents
* memory leak in dispd (cluster version dispathcer)
* wildcard pattern search in meta tags corrected
* memory leas in docd, affecting UPDATE and DELETE commands

=============================================================

3.2.5

* bugs fixed in SIMILAR command
* connection timeouts revised - works correctly in all situations