2000-11-27 Michael Widenius * Changed mi_create.c to use less stack. 2000-08-23 Michael Widenius * Fixed bug when comparing DECIMAL/NUMERIC key parts. 2000-08-17 Michael Widenius * Add a new flag in share.staus so that we can quickly check if a table is analyzed or not! 2000-07-02 Michael Widenius * Added safety margin to guard against full index file. 2000-05-22 Michael Widenius * Fixed that --join works with myisampack. 2000-05-14 Michael Widenius * Don't lock datafile during myisamchk (only indexfile is locked; This is good enough for all MyISAM functions); This made it possible to close datafile in rep_by_sort(). 2000-05-04 Michael Widenius * Fixed bug in code that scanned after rows in a crashed table. This could cause an infinite loop when repairing tables. 2000-04-26 Michael Widenius * Fixed bug when doing read_next after a delete/insert which balanced key pages (In this case one internal buffer was wrongly reused) 2000-04-21 Michael Widenius * Changed mi_find_halfpos() to return key, key_length and pos after key. * Don't join or split key buffers in the middle when inserting a key that is bigger than all other keys; This will improve inserts when doing these in sorted order. 2000-04-04 Michael Widenius * Added support for different languages on key part level. 2000-02-23 Michael Widenius * Fixed that myisamchk works properly with RAID. 2000-02-07 Michael Widenius * Added delete and rename of tables (works with RAID tables) 2000-01-29 Michael Widenius * Fixed the sorting of index works with prefix-packed keys. 1999-11-24 Michael Widenius * Fixed that DECIMAL() keys are sorted correct for negative numbers. 1999-11-22 Michael Widenius * removed 'NO_LOCKING' macros. * Added function mi_rnext_same * Added support for concurrent reads. 1999-11-05 Michael Widenius * Added function mi_scan(). * Changed all functions to return error number in case of errors. 1999-08-17 Michael Widenius * Added option DELAY_KEY_WRITE to tables and mi_open() 1999-08-10 Michael Widenius * Added support of HA_READ_PREFIX_LAST to mi_rkey(). This finds the last row with the given prefix. Mon Aug 2 13:54:35 1999 Michael Widenius * Added data- and key-file-length to myisamchk. * Fixed some problems with null and space packed keys. 1999-07-15 Michael Widenius * The following options are for COUNT(DISTINCT ..) * Added option HA_EXTRA_NO_ROWS; In this case only the index tree is updated * Added mi_delete_all_rows() 1999-07-13 Michael Widenius * Added special handling of tempoary tables 1999-06-12 Michael Widenius * Added optional checksum for file and for each dynamic-length row * Added unique constraint checking 1999-05-06 Michael Widenius * All index blocks of the same size now share the same key block delete link 1999-03-17 Michael Widenius * Different key packing code depending on if the first key part is a variable length column (space packed, BLOB or VARCHAR) * The create interface allows one to specify a key segment to start and end one a specific bit. (The bit handling isn't yet implemented) * Added more tests to 'test1' 1999-03-16 Michael Widenius * Added option -m to myisamchk as an alternative to -e (-m is faster but not as quite as safe as -e). * Added option --fast to not check not changed tables. * The first update will set a bit that the table has been changed. * The first update to a table increments a 'open_count' field. This will be reset on close. This will allow myisamchk to find tables that hasn't been properly closed! * Support for true VARCHAR columns 1999-03-01 Michael Widenius * Dynamic length blocks are double linked to allow easy reallocation of block lengths. This will help that the dynamic length data will not be as fragmented as with ISAM. * Extended mypack_isam to compress BLOB/TEXT columns. * Allow keys on BLOB. 1999-02-06 Michael Widenius * Keys, key pointers and all varibles in the index file are stored in high-endian-order to get better compression. * Allow NULL on keys 1998-10-29 Michael Widenius * All data is stored in low-endian order (This means that the files will be architecture and OS independent) * All record numbers are now of type 'ha_rows' and file pointer are now of type my_off_t. One can use files with 32-64 bit pointers with a 32bit or 64bit record handling. Currently the code is limited to 5 bytes pointers, but this is real easy to change.