0.8.0.3

Changes:

• disk read is by default verified by second read after 50ms to avoid consistency problems in some filesystems like ext4, btrfs

0.8.0.2

Changes:

• code cleanup
• clean up of database user grants
• transaction consistency bugs fixed
• added "schemaless mode"
• fixed batch mode startup

0.8.0

Fixed minor bugs.

0.8.0-beta7: Merge pull request #16 from bersler/nightly

Changes:

• fixed many bugs,
• add network stream output - in plain TCP/IP or ZeroMQ,
• rewritten transaction processing - now it is based on LWN,
• custom archived log files format.
• added support for database version 19.9 and 18.12
• added output format in Protobuf,
• rewritten checkpointing code,
• rewritten Kafka code for faster streaming - every sent message is independently confirmed,
• added support for idempotent Kafka output

0.7.0

Added support for reading redo log files from ASM instance (very slow, still experimental)

0.6.9

Added support for Big Endian architecture: Solaris Sparc
Added verification of existence of minimal supplemental logging.

0.6.5

Added support for 137 character, all possible on x86_64 platform.
Unified memory management, faster startup, memory is released when not needed

0.6.2

Added checks for database activation id
Schema information is stored to support offline mode

0.6

Added support for 4k block size redo log
Added support for multi block split for UNDO vectors
Added support for RAW type
Added support for TIMESTAMP WITH TIME ZONE type
Added support for invisible columns
Added support for columns with default value added when table had data
Many bugs fixed, code cleanup

0.5.14

Added CRC sum validation for read blocks
Added support for out of order UNDO blocks
Added mapping path for redo log files
Added checks of grants to system tables on start
Added supplemental log checks on start
Added support for custom primary key definition
Rewritten redo vector handling
Many bugs fixed