~drizzle-trunk/drizzle/development

/* Copyright (C) 2000-2006 MySQL AB

   This program is free software; you can redistribute it and/or modify

   it under the terms of the GNU General Public License as published by

   the Free Software Foundation; version 2 of the License.

   This program is distributed in the hope that it will be useful,

   but WITHOUT ANY WARRANTY; without even the implied warranty of

   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the

   GNU General Public License for more details.

   You should have received a copy of the GNU General Public License

   along with this program; if not, write to the Free Software

   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA */

/* Definitions for parameters to do with handler-routines */

#ifdef USE_PRAGMA_INTERFACE

#pragma interface			/* gcc class implementation */

#endif

#include <mysys/my_handler.h>

#include <storage/myisam/keycache.h>

#ifndef NO_HASH

#define NO_HASH				/* Not yet implemented */

#endif

// the following is for checking tables

#define HA_ADMIN_ALREADY_DONE	  1

#define HA_ADMIN_OK               0

#define HA_ADMIN_NOT_IMPLEMENTED -1

#define HA_ADMIN_FAILED		 -2

#define HA_ADMIN_CORRUPT         -3

#define HA_ADMIN_INTERNAL_ERROR  -4

#define HA_ADMIN_INVALID         -5

#define HA_ADMIN_REJECT          -6

#define HA_ADMIN_TRY_ALTER       -7

#define HA_ADMIN_WRONG_CHECKSUM  -8

#define HA_ADMIN_NOT_BASE_TABLE  -9

#define HA_ADMIN_NEEDS_UPGRADE  -10

#define HA_ADMIN_NEEDS_ALTER    -11

#define HA_ADMIN_NEEDS_CHECK    -12

/* Bits to show what an alter table will do */

#include <drizzled/sql_bitmap.h>

#define HA_MAX_ALTER_FLAGS 39

typedef Bitmap<HA_MAX_ALTER_FLAGS> HA_ALTER_FLAGS;

#define HA_ADD_INDEX                  (0)

#define HA_DROP_INDEX                 (1)

#define HA_ALTER_INDEX                (2)

#define HA_RENAME_INDEX               (3)

#define HA_ADD_UNIQUE_INDEX           (4)

#define HA_DROP_UNIQUE_INDEX          (5)

#define HA_ALTER_UNIQUE_INDEX         (6)

#define HA_RENAME_UNIQUE_INDEX        (7)

#define HA_ADD_PK_INDEX               (8)

#define HA_DROP_PK_INDEX              (9)

#define HA_ALTER_PK_INDEX             (10)

#define HA_ADD_COLUMN                 (11)

#define HA_DROP_COLUMN                (12)

#define HA_CHANGE_COLUMN              (13)

#define HA_ALTER_COLUMN_NAME          (14)

#define HA_ALTER_COLUMN_TYPE          (15)

#define HA_ALTER_COLUMN_ORDER         (16)

#define HA_ALTER_COLUMN_NULLABLE      (17)

#define HA_COLUMN_DEFAULT_VALUE       (18)

#define HA_COLUMN_STORAGE             (19)

#define HA_COLUMN_FORMAT              (20)

#define HA_ADD_FOREIGN_KEY            (21)

#define HA_DROP_FOREIGN_KEY           (22)

#define HA_ALTER_FOREIGN_KEY          (23)

#define HA_ADD_CONSTRAINT             (24)

#define HA_CHANGE_CHARACTER_SET       (30)

#define HA_SET_DEFAULT_CHARACTER_SET  (31)

#define HA_CHANGE_AUTOINCREMENT_VALUE (32)

#define HA_ALTER_STORAGE              (33)

#define HA_ALTER_TABLESPACE           (34)

#define HA_ALTER_ROW_FORMAT           (35)

#define HA_RENAME_TABLE               (36)

#define HA_ALTER_STORAGE_ENGINE       (37)

#define HA_RECREATE                   (38)

/* Remember to increase HA_MAX_ALTER_FLAGS when adding more flags! */

/* Return values for check_if_supported_alter */

#define HA_ALTER_ERROR               -1

#define HA_ALTER_SUPPORTED_WAIT_LOCK  0

#define HA_ALTER_SUPPORTED_NO_LOCK    1

#define HA_ALTER_NOT_SUPPORTED        2

/* Bits in table_flags() to show what database can do */

#define HA_NO_TRANSACTIONS     (1 << 0) /* Doesn't support transactions */

#define HA_PARTIAL_COLUMN_READ (1 << 1) /* read may not return all columns */

#define HA_TABLE_SCAN_ON_INDEX (1 << 2) /* No separate data/index file */

/*

  The following should be set if the following is not true when scanning

  a table with rnd_next()

  - We will see all rows (including deleted ones)

  - Row positions are 'table->s->db_record_offset' apart

  If this flag is not set, filesort will do a postion() call for each matched

  row to be able to find the row later.

*/

#define HA_REC_NOT_IN_SEQ      (1 << 3)

/* This is now a dead option, just left for compatibility */

#define HA_CAN_GEOMETRY        (1 << 4)

/*

  Reading keys in random order is as fast as reading keys in sort order

  (Used in records.cc to decide if we should use a record cache and by

  filesort to decide if we should sort key + data or key + pointer-to-row

*/

#define HA_FAST_KEY_READ       (1 << 5)

/*

  Set the following flag if we on delete should force all key to be read

  and on update read all keys that changes

*/

#define HA_REQUIRES_KEY_COLUMNS_FOR_DELETE (1 << 6)

#define HA_NULL_IN_KEY         (1 << 7) /* One can have keys with NULL */

#define HA_DUPLICATE_POS       (1 << 8)    /* ha_position() gives dup row */

#define HA_NO_BLOBS            (1 << 9) /* Doesn't support blobs */

#define HA_CAN_INDEX_BLOBS     (1 << 10)

#define HA_AUTO_PART_KEY       (1 << 11) /* auto-increment in multi-part key */

#define HA_REQUIRE_PRIMARY_KEY (1 << 12) /* .. and can't create a hidden one */

#define HA_STATS_RECORDS_IS_EXACT (1 << 13) /* stats.records is exact */

/*

  INSERT_DELAYED only works with handlers that uses MySQL internal table

  level locks

*/

#define HA_CAN_INSERT_DELAYED  (1 << 14)

/*

  If we get the primary key columns for free when we do an index read

  It also implies that we have to retrive the primary key when using

  position() and rnd_pos().

*/

#define HA_PRIMARY_KEY_IN_READ_INDEX (1 << 15)

/*

  If HA_PRIMARY_KEY_REQUIRED_FOR_POSITION is set, it means that to position()

  uses a primary key. Without primary key, we can't call position().

*/ 

#define HA_PRIMARY_KEY_REQUIRED_FOR_POSITION (1 << 16) 

#define HA_CAN_RTREEKEYS       (1 << 17) /* Historical, no longer supported */

#define HA_NOT_DELETE_WITH_CACHE (1 << 18)

/*

  The following is we need to a primary key to delete (and update) a row.

  If there is no primary key, all columns needs to be read on update and delete

*/

#define HA_PRIMARY_KEY_REQUIRED_FOR_DELETE (1 << 19)

#define HA_NO_PREFIX_CHAR_KEYS (1 << 20)

#define HA_CAN_FULLTEXT        (1 << 21) /* Historical, no longer supported */

#define HA_CAN_SQL_HANDLER     (1 << 22) /* Historical, no longer supported */

#define HA_NO_AUTO_INCREMENT   (1 << 23)

#define HA_HAS_CHECKSUM        (1 << 24)

/* Table data are stored in separate files (for lower_case_table_names) */

#define HA_FILE_BASED	       (1 << 26)

#define HA_NO_VARCHAR	       (1 << 27) /* Historical, no longer supported */

#define HA_CAN_BIT_FIELD       (1 << 28) /* Historical, no longer supported */

#define HA_NEED_READ_RANGE_BUFFER (1 << 29) /* for read_multi_range */

#define HA_ANY_INDEX_MAY_BE_UNIQUE (1 << 30)

#define HA_NO_COPY_ON_ALTER    (1LL << 31)

#define HA_HAS_RECORDS	       (1LL << 32) /* records() gives exact count*/

/* Has it's own method of binlog logging */

#define HA_HAS_OWN_BINLOGGING  (1LL << 33) /* Historical, no longer supported */

#define HA_MRR_CANT_SORT       (1LL << 34)

/*

  Engine is capable of row-format and statement-format logging,

  respectively

*/

#define HA_BINLOG_ROW_CAPABLE  (1LL << 35)

#define HA_BINLOG_STMT_CAPABLE (1LL << 36)

#define HA_ONLINE_ALTER        (1LL << 37)

/*

  Set of all binlog flags. Currently only contain the capabilities

  flags.

 */

#define HA_BINLOG_FLAGS (HA_BINLOG_ROW_CAPABLE | HA_BINLOG_STMT_CAPABLE)

/* bits in index_flags(index_number) for what you can do with index */

#define HA_READ_NEXT            1       /* TODO really use this flag */

#define HA_READ_PREV            2       /* supports ::index_prev */

#define HA_READ_ORDER           4       /* index_next/prev follow sort order */

#define HA_READ_RANGE           8       /* can find all records in a range */

#define HA_ONLY_WHOLE_INDEX	16	/* Can't use part key searches */

#define HA_KEYREAD_ONLY         64	/* Support HA_EXTRA_KEYREAD */

/*

  Index scan will not return records in rowid order. Not guaranteed to be

  set for unordered (e.g. HASH) indexes.

*/

#define HA_KEY_SCAN_NOT_ROR     128 

#define HA_DO_INDEX_COND_PUSHDOWN  256 /* Supports Index Condition Pushdown */

/*

  HA_PARTITION_FUNCTION_SUPPORTED indicates that the function is

  supported at all.

  HA_FAST_CHANGE_PARTITION means that optimised variants of the changes

  exists but they are not necessarily done online.

  HA_ONLINE_DOUBLE_WRITE means that the handler supports writing to both

  the new partition and to the old partitions when updating through the

  old partitioning schema while performing a change of the partitioning.

  This means that we can support updating of the table while performing

  the copy phase of the change. For no lock at all also a double write

  from new to old must exist and this is not required when this flag is

  set.

  This is actually removed even before it was introduced the first time.

  The new idea is that handlers will handle the lock level already in

  store_lock for ALTER TABLE partitions.

  HA_PARTITION_ONE_PHASE is a flag that can be set by handlers that take

  care of changing the partitions online and in one phase. Thus all phases

  needed to handle the change are implemented inside the storage engine.

  The storage engine must also support auto-discovery since the frm file

  is changed as part of the change and this change must be controlled by

  the storage engine. A typical engine to support this is NDB (through

  WL #2498).

*/

#define HA_PARTITION_FUNCTION_SUPPORTED         (1L << 1)

#define HA_FAST_CHANGE_PARTITION                (1L << 2)

#define HA_PARTITION_ONE_PHASE                  (1L << 3)

/* operations for disable/enable indexes */

#define HA_KEY_SWITCH_NONUNIQ      0

#define HA_KEY_SWITCH_ALL          1

#define HA_KEY_SWITCH_NONUNIQ_SAVE 2

#define HA_KEY_SWITCH_ALL_SAVE     3

/*

  Note: the following includes binlog and closing 0.

  so: innodb + bdb + ndb + binlog + myisam + myisammrg + archive +

      example + csv + heap + blackhole + federated + 0

  (yes, the sum is deliberately inaccurate)

  TODO remove the limit, use dynarrays

*/

#define MAX_HA 15

/*

  Parameters for open() (in register form->filestat)

  HA_GET_INFO does an implicit HA_ABORT_IF_LOCKED

*/

#define HA_OPEN_KEYFILE		1

#define HA_OPEN_RNDFILE		2

#define HA_GET_INDEX		4

#define HA_GET_INFO		8	/* do a ha_info() after open */

#define HA_READ_ONLY		16	/* File opened as readonly */

/* Try readonly if can't open with read and write */

#define HA_TRY_READ_ONLY	32

#define HA_WAIT_IF_LOCKED	64	/* Wait if locked on open */

#define HA_ABORT_IF_LOCKED	128	/* skip if locked on open.*/

#define HA_BLOCK_LOCK		256	/* unlock when reading some records */

#define HA_OPEN_TEMPORARY	512

/* For transactional LOCK TABLE. handler::lock_table() */

#define HA_LOCK_IN_SHARE_MODE      F_RDLCK

#define HA_LOCK_IN_EXCLUSIVE_MODE  F_WRLCK

/* Some key definitions */

#define HA_KEY_NULL_LENGTH	1

#define HA_KEY_BLOB_LENGTH	2

#define HA_LEX_CREATE_TMP_TABLE	1

#define HA_LEX_CREATE_IF_NOT_EXISTS 2

#define HA_LEX_CREATE_TABLE_LIKE 4

#define HA_OPTION_NO_CHECKSUM	(1L << 17)

#define HA_OPTION_NO_DELAY_KEY_WRITE (1L << 18)

#define HA_MAX_REC_LENGTH	65535

/* Table caching type */

#define HA_CACHE_TBL_NONTRANSACT 0

#define HA_CACHE_TBL_NOCACHE     1

#define HA_CACHE_TBL_ASKTRANSACT 2

#define HA_CACHE_TBL_TRANSACT    4

/* Options of START TRANSACTION statement (and later of SET TRANSACTION stmt) */

#define MYSQL_START_TRANS_OPT_WITH_CONS_SNAPSHOT 1

/* Flags for method is_fatal_error */

#define HA_CHECK_DUP_KEY 1

#define HA_CHECK_DUP_UNIQUE 2

#define HA_CHECK_DUP (HA_CHECK_DUP_KEY + HA_CHECK_DUP_UNIQUE)

enum legacy_db_type

{

  DB_TYPE_UNKNOWN=0,DB_TYPE_DIAB_ISAM=1,

  DB_TYPE_HASH,DB_TYPE_MISAM,DB_TYPE_PISAM,

  DB_TYPE_RMS_ISAM, DB_TYPE_HEAP, DB_TYPE_ISAM,

  DB_TYPE_MRG_ISAM, DB_TYPE_MYISAM, DB_TYPE_MRG_MYISAM,

  DB_TYPE_BERKELEY_DB, DB_TYPE_INNODB,

  DB_TYPE_GEMINI, DB_TYPE_NDBCLUSTER,

  DB_TYPE_EXAMPLE_DB, DB_TYPE_ARCHIVE_DB, DB_TYPE_CSV_DB,

  DB_TYPE_FEDERATED_DB,

  DB_TYPE_BLACKHOLE_DB,

  DB_TYPE_PARTITION_DB,

  DB_TYPE_BINLOG,

  DB_TYPE_SOLID,

  DB_TYPE_PBXT,

  DB_TYPE_TABLE_FUNCTION,

  DB_TYPE_MEMCACHE,

  DB_TYPE_FALCON,

  DB_TYPE_MARIA,

  DB_TYPE_FIRST_DYNAMIC=42,

  DB_TYPE_DEFAULT=127 // Must be last

};

enum row_type { ROW_TYPE_NOT_USED=-1, ROW_TYPE_DEFAULT, ROW_TYPE_FIXED,

		ROW_TYPE_DYNAMIC, ROW_TYPE_COMPRESSED,

		ROW_TYPE_REDUNDANT, ROW_TYPE_COMPACT, ROW_TYPE_PAGE };

enum column_format_type { COLUMN_FORMAT_TYPE_NOT_USED= -1,

                          COLUMN_FORMAT_TYPE_DEFAULT=   0,

                          COLUMN_FORMAT_TYPE_FIXED=     1,

                          COLUMN_FORMAT_TYPE_DYNAMIC=   2 };

enum enum_binlog_func {

  BFN_RESET_LOGS=        1,

  BFN_RESET_SLAVE=       2,

  BFN_BINLOG_WAIT=       3,

  BFN_BINLOG_END=        4,

  BFN_BINLOG_PURGE_FILE= 5

};

enum enum_binlog_command {

  LOGCOM_CREATE_TABLE,

  LOGCOM_ALTER_TABLE,

  LOGCOM_RENAME_TABLE,

  LOGCOM_DROP_TABLE,

  LOGCOM_CREATE_DB,

  LOGCOM_ALTER_DB,

  LOGCOM_DROP_DB

};

/* struct to hold information about the table that should be created */

/* Bits in used_fields */

#define HA_CREATE_USED_AUTO             (1L << 0)

#ifdef DEAD_OPTIONS

#define HA_CREATE_USED_RAID             (1L << 1) /* Historical, no longer supported */

#define HA_CREATE_USED_UNION            (1L << 2)

#define HA_CREATE_USED_PASSWORD         (1L << 17)

#endif

#define HA_CREATE_USED_INSERT_METHOD    (1L << 3)

#define HA_CREATE_USED_MIN_ROWS         (1L << 4)

#define HA_CREATE_USED_MAX_ROWS         (1L << 5)

#define HA_CREATE_USED_AVG_ROW_LENGTH   (1L << 6)

#define HA_CREATE_USED_PACK_KEYS        (1L << 7)

#define HA_CREATE_USED_CHARSET          (1L << 8)

#define HA_CREATE_USED_DEFAULT_CHARSET  (1L << 9)

#define HA_CREATE_USED_DATADIR          (1L << 10)

#define HA_CREATE_USED_INDEXDIR         (1L << 11)

#define HA_CREATE_USED_ENGINE           (1L << 12)

#define HA_CREATE_USED_CHECKSUM         (1L << 13)

#define HA_CREATE_USED_DELAY_KEY_WRITE  (1L << 14)

#define HA_CREATE_USED_ROW_FORMAT       (1L << 15)

#define HA_CREATE_USED_COMMENT          (1L << 16)

#define HA_CREATE_USED_CONNECTION       (1L << 18)

#define HA_CREATE_USED_KEY_BLOCK_SIZE   (1L << 19)

#define HA_CREATE_USED_TRANSACTIONAL    (1L << 20)

#define HA_CREATE_USED_PAGE_CHECKSUM    (1L << 21)

#define HA_CREATE_USED_BLOCK_SIZE       (1L << 22)

typedef uint64_t my_xid; // this line is the same as in log_event.h

#define MYSQL_XID_PREFIX "MySQLXid"

#define MYSQL_XID_PREFIX_LEN 8 // must be a multiple of 8

#define MYSQL_XID_OFFSET (MYSQL_XID_PREFIX_LEN+sizeof(server_id))

#define MYSQL_XID_GTRID_LEN (MYSQL_XID_OFFSET+sizeof(my_xid))

#define XIDDATASIZE MYSQL_XIDDATASIZE

#define MAXGTRIDSIZE 64

#define MAXBQUALSIZE 64

#define COMPATIBLE_DATA_YES 0

#define COMPATIBLE_DATA_NO  1

typedef bool (*qc_engine_callback)(THD *thd, char *table_key,

                                      uint key_length,

                                      uint64_t *engine_data);

/**

  struct xid_t is binary compatible with the XID structure as

  in the X/Open CAE Specification, Distributed Transaction Processing:

  The XA Specification, X/Open Company Ltd., 1991.

  http://www.opengroup.org/bookstore/catalog/c193.htm

  @see MYSQL_XID in mysql/plugin.h

*/

struct xid_t {

  long formatID;

  long gtrid_length;

  long bqual_length;

  char data[XIDDATASIZE];  // not \0-terminated !

  xid_t() {}                                /* Remove gcc warning */  

  bool eq(struct xid_t *xid)

  { return eq(xid->gtrid_length, xid->bqual_length, xid->data); }

  bool eq(long g, long b, const char *d)

  { return g == gtrid_length && b == bqual_length && !memcmp(d, data, g+b); }

  void set(struct xid_t *xid)

  { memcpy(this, xid, xid->length()); }

  void set(long f, const char *g, long gl, const char *b, long bl)

  {

    formatID= f;

    memcpy(data, g, gtrid_length= gl);

    memcpy(data+gl, b, bqual_length= bl);

  }

  void set(uint64_t xid)

  {

    my_xid tmp;

    formatID= 1;

    set(MYSQL_XID_PREFIX_LEN, 0, MYSQL_XID_PREFIX);

    memcpy(data+MYSQL_XID_PREFIX_LEN, &server_id, sizeof(server_id));

    tmp= xid;

    memcpy(data+MYSQL_XID_OFFSET, &tmp, sizeof(tmp));

    gtrid_length=MYSQL_XID_GTRID_LEN;

  }

  void set(long g, long b, const char *d)

  {

    formatID= 1;

    gtrid_length= g;

    bqual_length= b;

    memcpy(data, d, g+b);

  }

  bool is_null() { return formatID == -1; }

  void null() { formatID= -1; }

  my_xid quick_get_my_xid()

  {

    my_xid tmp;

    memcpy(&tmp, data+MYSQL_XID_OFFSET, sizeof(tmp));

    return tmp;

  }

  my_xid get_my_xid()

  {

    return gtrid_length == MYSQL_XID_GTRID_LEN && bqual_length == 0 &&

           !memcmp(data+MYSQL_XID_PREFIX_LEN, &server_id, sizeof(server_id)) &&

           !memcmp(data, MYSQL_XID_PREFIX, MYSQL_XID_PREFIX_LEN) ?

           quick_get_my_xid() : 0;

  }

  uint length()

  {

    return sizeof(formatID)+sizeof(gtrid_length)+sizeof(bqual_length)+

           gtrid_length+bqual_length;

  }

  uchar *key()

  {

    return (uchar *)&gtrid_length;

  }

  uint key_length()

  {

    return sizeof(gtrid_length)+sizeof(bqual_length)+gtrid_length+bqual_length;

  }

};

typedef struct xid_t XID;

/* for recover() handlerton call */

#define MIN_XID_LIST_SIZE  128

#define MAX_XID_LIST_SIZE  (1024*128)

struct handlerton;

/* The handler for a table type.  Will be included in the TABLE structure */

struct st_table;

typedef struct st_table TABLE;

typedef struct st_table_share TABLE_SHARE;

struct st_foreign_key_info;

typedef struct st_foreign_key_info FOREIGN_KEY_INFO;

typedef bool (stat_print_fn)(THD *thd, const char *type, uint type_len,

                             const char *file, uint file_len,

                             const char *status, uint status_len);

enum ha_stat_type { HA_ENGINE_STATUS, HA_ENGINE_LOGS, HA_ENGINE_MUTEX };

extern st_plugin_int *hton2plugin[MAX_HA];

/*

  handlerton is a singleton structure - one instance per storage engine -

  to provide access to storage engine functionality that works on the

  "global" level (unlike handler class that works on a per-table basis)

  usually handlerton instance is defined statically in ha_xxx.cc as

  static handlerton { ... } xxx_hton;

  savepoint_*, prepare, recover, and *_by_xid pointers can be 0.

*/

struct handlerton

{

  /*

    Historical marker for if the engine is available of not

  */

  SHOW_COMP_OPTION state;

  /*

    Historical number used for frm file to determine the correct storage engine.

    This is going away and new engines will just use "name" for this.

  */

  enum legacy_db_type db_type;

  /*

    each storage engine has it's own memory area (actually a pointer)

    in the thd, for storing per-connection information.

    It is accessed as

      thd->ha_data[xxx_hton.slot]

   slot number is initialized by MySQL after xxx_init() is called.

   */

   uint slot;

   /*

     to store per-savepoint data storage engine is provided with an area

     of a requested size (0 is ok here).

     savepoint_offset must be initialized statically to the size of

     the needed memory to store per-savepoint information.

     After xxx_init it is changed to be an offset to savepoint storage

     area and need not be used by storage engine.

     see binlog_hton and binlog_savepoint_set/rollback for an example.

   */

   uint savepoint_offset;

   /*

     handlerton methods:

     close_connection is only called if

     thd->ha_data[xxx_hton.slot] is non-zero, so even if you don't need

     this storage area - set it to something, so that MySQL would know

     this storage engine was accessed in this connection

   */

   int  (*close_connection)(handlerton *hton, THD *thd);

   /*

     sv points to an uninitialized storage area of requested size

     (see savepoint_offset description)

   */

   int  (*savepoint_set)(handlerton *hton, THD *thd, void *sv);

   /*

     sv points to a storage area, that was earlier passed

     to the savepoint_set call

   */

   int  (*savepoint_rollback)(handlerton *hton, THD *thd, void *sv);

   int  (*savepoint_release)(handlerton *hton, THD *thd, void *sv);

   /*

     'all' is true if it's a real commit, that makes persistent changes

     'all' is false if it's not in fact a commit but an end of the

     statement that is part of the transaction.

     NOTE 'all' is also false in auto-commit mode where 'end of statement'

     and 'real commit' mean the same event.

   */

   int  (*commit)(handlerton *hton, THD *thd, bool all);

   int  (*rollback)(handlerton *hton, THD *thd, bool all);

   int  (*prepare)(handlerton *hton, THD *thd, bool all);

   int  (*recover)(handlerton *hton, XID *xid_list, uint len);

   int  (*commit_by_xid)(handlerton *hton, XID *xid);

   int  (*rollback_by_xid)(handlerton *hton, XID *xid);

   void *(*create_cursor_read_view)(handlerton *hton, THD *thd);

   void (*set_cursor_read_view)(handlerton *hton, THD *thd, void *read_view);

   void (*close_cursor_read_view)(handlerton *hton, THD *thd, void *read_view);

   handler *(*create)(handlerton *hton, TABLE_SHARE *table, MEM_ROOT *mem_root);

   void (*drop_database)(handlerton *hton, char* path);

   int (*start_consistent_snapshot)(handlerton *hton, THD *thd);

   bool (*flush_logs)(handlerton *hton);

   bool (*show_status)(handlerton *hton, THD *thd, stat_print_fn *print, enum ha_stat_type stat);

   int (*fill_files_table)(handlerton *hton, THD *thd,

                           TABLE_LIST *tables,

                           class Item *cond);

   uint32_t flags;                                /* global handler flags */

   int (*release_temporary_latches)(handlerton *hton, THD *thd);

   int (*discover)(handlerton *hton, THD* thd, const char *db, 

                   const char *name,

                   uchar **frmblob, 

                   size_t *frmlen);

   int (*table_exists_in_engine)(handlerton *hton, THD* thd, const char *db,

                                 const char *name);

   uint32_t license; /* Flag for Engine License */

   void *data; /* Location for engines to keep personal structures */

};

/* Possible flags of a handlerton (there can be 32 of them) */

#define HTON_NO_FLAGS                 0

#define HTON_CLOSE_CURSORS_AT_COMMIT (1 << 0)

#define HTON_ALTER_NOT_SUPPORTED     (1 << 1) //Engine does not support alter

#define HTON_CAN_RECREATE            (1 << 2) //Delete all is used fro truncate

#define HTON_HIDDEN                  (1 << 3) //Engine does not appear in lists

#define HTON_FLUSH_AFTER_RENAME      (1 << 4)

#define HTON_NOT_USER_SELECTABLE     (1 << 5)

#define HTON_TEMPORARY_NOT_SUPPORTED (1 << 6) //Having temporary tables not supported

#define HTON_SUPPORT_LOG_TABLES      (1 << 7) //Engine supports log tables

#define HTON_NO_PARTITION            (1 << 8) //You can not partition these tables

class Ha_trx_info;

struct THD_TRANS

{

  /* true is not all entries in the ht[] support 2pc */

  bool        no_2pc;

  /* storage engines that registered in this transaction */

  Ha_trx_info *ha_list;

  /* 

    The purpose of this flag is to keep track of non-transactional

    tables that were modified in scope of:

    - transaction, when the variable is a member of

    THD::transaction.all

    - top-level statement or sub-statement, when the variable is a

    member of THD::transaction.stmt

    This member has the following life cycle:

    * stmt.modified_non_trans_table is used to keep track of

    modified non-transactional tables of top-level statements. At

    the end of the previous statement and at the beginning of the session,

    it is reset to false.  If such functions

    as mysql_insert, mysql_update, mysql_delete etc modify a

    non-transactional table, they set this flag to true.  At the

    end of the statement, the value of stmt.modified_non_trans_table 

    is merged with all.modified_non_trans_table and gets reset.

    * all.modified_non_trans_table is reset at the end of transaction

    * Since we do not have a dedicated context for execution of a

    sub-statement, to keep track of non-transactional changes in a

    sub-statement, we re-use stmt.modified_non_trans_table. 

    At entrance into a sub-statement, a copy of the value of

    stmt.modified_non_trans_table (containing the changes of the

    outer statement) is saved on stack. Then 

    stmt.modified_non_trans_table is reset to false and the

    substatement is executed. Then the new value is merged with the

    saved value.

  */

  bool modified_non_trans_table;

  void reset() { no_2pc= false; modified_non_trans_table= false; }

};

/**

  Either statement transaction or normal transaction - related

  thread-specific storage engine data.

  If a storage engine participates in a statement/transaction,

  an instance of this class is present in

  thd->transaction.{stmt|all}.ha_list. The addition to

  {stmt|all}.ha_list is made by trans_register_ha().

  When it's time to commit or rollback, each element of ha_list

  is used to access storage engine's prepare()/commit()/rollback()

  methods, and also to evaluate if a full two phase commit is

  necessary.

  @sa General description of transaction handling in handler.cc.

*/

class Ha_trx_info

{

public:

  /** Register this storage engine in the given transaction context. */

  void register_ha(THD_TRANS *trans, handlerton *ht_arg)

  {

    assert(m_flags == 0);

    assert(m_ht == NULL);

    assert(m_next == NULL);

    m_ht= ht_arg;

    m_flags= (int) TRX_READ_ONLY; /* Assume read-only at start. */

    m_next= trans->ha_list;

    trans->ha_list= this;

  }

  /** Clear, prepare for reuse. */

  void reset()

  {

    m_next= NULL;

    m_ht= NULL;

    m_flags= 0;

  }

  Ha_trx_info() { reset(); }

  void set_trx_read_write()

  {

    assert(is_started());

    m_flags|= (int) TRX_READ_WRITE;

  }

  bool is_trx_read_write() const

  {

    assert(is_started());

    return m_flags & (int) TRX_READ_WRITE;

  }

  bool is_started() const { return m_ht != NULL; }

  /** Mark this transaction read-write if the argument is read-write. */

  void coalesce_trx_with(const Ha_trx_info *stmt_trx)

  {

    /*

      Must be called only after the transaction has been started.

      Can be called many times, e.g. when we have many

      read-write statements in a transaction.

    */

    assert(is_started());

    if (stmt_trx->is_trx_read_write())

      set_trx_read_write();

  }

  Ha_trx_info *next() const

  {

    assert(is_started());

    return m_next;

  }

  handlerton *ht() const

  {

    assert(is_started());

    return m_ht;

  }

private:

  enum { TRX_READ_ONLY= 0, TRX_READ_WRITE= 1 };

  /** Auxiliary, used for ha_list management */

  Ha_trx_info *m_next;

  /**

    Although a given Ha_trx_info instance is currently always used

    for the same storage engine, 'ht' is not-NULL only when the

    corresponding storage is a part of a transaction.

  */

  handlerton *m_ht;

  /**

    Transaction flags related to this engine.

    Not-null only if this instance is a part of transaction.

    May assume a combination of enum values above.

  */

  uchar       m_flags;

};

enum enum_tx_isolation { ISO_READ_UNCOMMITTED, ISO_READ_COMMITTED,

                         ISO_REPEATABLE_READ, ISO_SERIALIZABLE};

typedef struct {

  uint64_t data_file_length;

  uint64_t max_data_file_length;

  uint64_t index_file_length;

  uint64_t delete_length;

  ha_rows records;

  uint32_t mean_rec_length;

  time_t create_time;

  time_t check_time;

  time_t update_time;

  uint64_t check_sum;

} PARTITION_INFO;

#define UNDEF_NODEGROUP 65535

class Item;

struct st_table_log_memory_entry;

#define NOT_A_PARTITION_ID ((uint32_t)-1)

enum ha_choice { HA_CHOICE_UNDEF, HA_CHOICE_NO, HA_CHOICE_YES };

typedef struct st_ha_create_information

{

  const CHARSET_INFO *table_charset, *default_table_charset;

  LEX_STRING connect_string;

  LEX_STRING comment;

  const char *data_file_name, *index_file_name;

  const char *alias;

  uint64_t max_rows,min_rows;

  uint64_t auto_increment_value;

  uint32_t table_options;

  uint32_t avg_row_length;

  uint32_t used_fields;

  uint32_t key_block_size;

  uint32_t block_size;

  SQL_LIST merge_list;

  handlerton *db_type;

  enum row_type row_type;

  uint null_bits;                       /* NULL bits at start of record */

  uint options;				/* OR of HA_CREATE_ options */

  uint merge_insert_method;

  uint extra_size;                      /* length of extra data segment */

  /* 0 not used, 1 if not transactional, 2 if transactional */

  enum ha_choice transactional;

  bool table_existed;			/* 1 in create if table existed */

  bool frm_only;                        /* 1 if no ha_create_table() */

  bool varchar;                         /* 1 if table has a VARCHAR */

  enum ha_choice page_checksum;         /* If we have page_checksums */

} HA_CREATE_INFO;

typedef struct st_ha_alter_information

{

  KEY  *key_info_buffer;

  uint key_count;

  uint index_drop_count;

  uint *index_drop_buffer;

  uint index_add_count;

  uint *index_add_buffer;

  void *data;

} HA_ALTER_INFO;

typedef struct st_key_create_information

{

  enum ha_key_alg algorithm;

  uint32_t block_size;

  LEX_STRING parser_name;

  LEX_STRING comment;

} KEY_CREATE_INFO;

/*

  Class for maintaining hooks used inside operations on tables such

  as: create table functions, delete table functions, and alter table

  functions.

  Class is using the Template Method pattern to separate the public

  usage interface from the private inheritance interface.  This

  imposes no overhead, since the public non-virtual function is small

  enough to be inlined.

  The hooks are usually used for functions that does several things,

  e.g., create_table_from_items(), which both create a table and lock

  it.

 */

class TABLEOP_HOOKS

{

public:

  TABLEOP_HOOKS() {}

  virtual ~TABLEOP_HOOKS() {}

  inline void prelock(TABLE **tables, uint count)

  {

    do_prelock(tables, count);

  }

  inline int postlock(TABLE **tables, uint count)

  {

    return do_postlock(tables, count);

  }

private:

  /* Function primitive that is called prior to locking tables */

  virtual void do_prelock(TABLE **tables __attribute__((unused)),

                          uint count __attribute__((unused)))

  {

    /* Default is to do nothing */

  }

  /**

     Primitive called after tables are locked.

     If an error is returned, the tables will be unlocked and error

     handling start.

     @return Error code or zero.

   */

  virtual int do_postlock(TABLE **tables __attribute__((unused)),

                          uint count __attribute__((unused)))

  {

    return 0;                           /* Default is to do nothing */

  }

};

typedef struct st_savepoint SAVEPOINT;

extern uint32_t savepoint_alloc_size;

extern KEY_CREATE_INFO default_key_create_info;

/* Forward declaration for condition pushdown to storage engine */

typedef class Item COND;

typedef struct st_ha_check_opt

{

  st_ha_check_opt() {}                        /* Remove gcc warning */

  uint32_t sort_buffer_size;

  uint flags;       /* isam layer flags (e.g. for myisamchk) */

  uint sql_flags;   /* sql layer flags - for something myisamchk cannot do */

  KEY_CACHE *key_cache;	/* new key cache when changing key cache */

  void init();

} HA_CHECK_OPT;

/*

  This is a buffer area that the handler can use to store rows.

  'end_of_used_area' should be kept updated after calls to

  read-functions so that other parts of the code can use the

  remaining area (until next read calls is issued).

*/

typedef struct st_handler_buffer

{

  uchar *buffer;         /* Buffer one can start using */

  uchar *buffer_end;     /* End of buffer */

  uchar *end_of_used_area;     /* End of area that was used by handler */

} HANDLER_BUFFER;

typedef struct system_status_var SSV;

typedef void *range_seq_t;

typedef struct st_range_seq_if

{

  /*