1.1 (2013-11-07)

Bug fix:

• Panic due to out of memory on several 32-bit machines.
• Fix several incorrect HTTP API content type.
• Fix wrong new bucket position in opened hash tables.

Improvements:

• Remove per-collection padding buffer and replace it by a shared string buffer, to reduce memory consumption.
• Creation of data file no longer creates a giant empty buffer beforehand, therefore reducing memory consumption.
• Documents may now have optional persistent IDs (called UID) which will never change during its life time.
• API version 3 (New and backward compatible) supports document operations based on UIDs.
• Lock granularity is further tweaked.
• When a document cannot be indexed due to having incompatible data structure, a warning message is logged.
• Query now supports regex collection scan and reversed integer range lookup.

1.0 (2013-09-21)

A maintenance release to address all outstanding issues found in alpha and beta, with feature improvements and new APIs.

Bug fix:

• Scalability problem on a model of laptop has been resolved.
• Collection update will no longer panic under a rare data corruption situation.

Improvements:

• Data file IO now uses more granular locks (RWlock-per-bucket and RWlock-per-document) instead of giant RW file lock.
• API version 2 (New and backward compatible)
• New HTTP endpoints to report server runtime performance.
• New query syntax - easier and more efficient, together with a new query processor.
• A specific type of range query (integer lookup in a range) is now supported.