~drizzle-trunk/drizzle/development

/* -*- mode: c++; c-basic-offset: 2; indent-tabs-mode: nil; -*-
 *  vim:expandtab:shiftwidth=2:tabstop=2:smarttab:
 *
 *  Copyright (C) 2008 Sun Microsystems, Inc.
 *
 *  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., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
 */



#ifndef DRIZZLED_SESSION_H
#define DRIZZLED_SESSION_H

#include "drizzled/plugin.h"
#include "drizzled/sql_locale.h"
#include "drizzled/resource_context.h"
#include "drizzled/cursor.h"
#include "drizzled/current_session.h"
#include "drizzled/sql_error.h"
#include "drizzled/file_exchange.h"
#include "drizzled/select_result_interceptor.h"
#include "drizzled/statistics_variables.h"
#include "drizzled/xid.h"
#include "drizzled/query_id.h"
#include "drizzled/named_savepoint.h"
#include "drizzled/transaction_context.h"
#include "drizzled/util/storable.h"
#include "drizzled/my_hash.h"
#include "drizzled/pthread_globals.h"
#include <netdb.h>
#include <sys/time.h>
#include <sys/resource.h>

#include <algorithm>
#include <bitset>
#include <deque>
#include <map>
#include <string>

#include "drizzled/identifier.h"
#include "drizzled/open_tables_state.h"
#include "drizzled/internal_error_handler.h"
#include "drizzled/diagnostics_area.h"
#include "drizzled/plugin/authorization.h"

#include "drizzled/catalog/instance.h"
#include "drizzled/catalog/local.h"

#include <boost/unordered_map.hpp>

#include <boost/thread/thread.hpp>
#include <boost/thread/mutex.hpp>
#include <boost/thread/shared_mutex.hpp>
#include <boost/thread/condition_variable.hpp>
#include <boost/make_shared.hpp>

#include "drizzled/visibility.h"

#define MIN_HANDSHAKE_SIZE      6

namespace drizzled
{

namespace plugin
{
class Client;
class Scheduler;
class EventObserverList;
}

namespace message
{
class Transaction;
class Statement;
class Resultset;
}

namespace internal
{
struct st_my_thread_var;
}

namespace table
{
class Placeholder;
}

class Lex_input_stream;
class user_var_entry;
class CopyField;
class Table_ident;

class TableShareInstance;

extern char internal_table_name[2];
extern char empty_c_string[1];
extern const char **errmesg;

#define TC_HEURISTIC_RECOVER_COMMIT   1
#define TC_HEURISTIC_RECOVER_ROLLBACK 2
extern uint32_t tc_heuristic_recover;

/**
  @brief
  Local storage for proto that are tmp table. This should be enlarged
  to hande the entire table-share for a local table. Once Hash is done,
  we should consider exchanging the map for it.
*/
typedef std::map <std::string, message::Table> ProtoCache;

/**
  The COPY_INFO structure is used by INSERT/REPLACE code.
  The schema of the row counting by the INSERT/INSERT ... ON DUPLICATE KEY
  UPDATE code:
    If a row is inserted then the copied variable is incremented.
    If a row is updated by the INSERT ... ON DUPLICATE KEY UPDATE and the
      new data differs from the old one then the copied and the updated
      variables are incremented.
    The touched variable is incremented if a row was touched by the update part
      of the INSERT ... ON DUPLICATE KEY UPDATE no matter whether the row
      was actually changed or not.
*/
class CopyInfo 
{
public:
  ha_rows records; /**< Number of processed records */
  ha_rows deleted; /**< Number of deleted records */
  ha_rows updated; /**< Number of updated records */
  ha_rows copied;  /**< Number of copied records */
  ha_rows error_count;
  ha_rows touched; /* Number of touched records */
  enum enum_duplicates handle_duplicates;
  int escape_char, last_errno;
  bool ignore;
  /* for INSERT ... UPDATE */
  List<Item> *update_fields;
  List<Item> *update_values;
  /* for VIEW ... WITH CHECK OPTION */

  CopyInfo() :
    records(0),
    deleted(0),
    updated(0),
    copied(0),
    error_count(0),
    touched(0),
    escape_char(0),
    last_errno(0),
    ignore(0),
    update_fields(0),
    update_values(0)
  { }

};

} /* namespace drizzled */

/** @TODO why is this in the middle of the file */
#include <drizzled/lex_column.h>

namespace drizzled
{

class select_result;
class Time_zone;

#define Session_SENTRY_MAGIC 0xfeedd1ff
#define Session_SENTRY_GONE  0xdeadbeef

struct drizzle_system_variables
{
  drizzle_system_variables()
  {}
  /*
    How dynamically allocated system variables are handled:

    The global_system_variables and max_system_variables are "authoritative"
    They both should have the same 'version' and 'size'.
    When attempting to access a dynamic variable, if the session version
    is out of date, then the session version is updated and realloced if
    neccessary and bytes copied from global to make up for missing data.
  */
  ulong dynamic_variables_version;
  char * dynamic_variables_ptr;
  uint32_t dynamic_variables_head;  /* largest valid variable offset */
  uint32_t dynamic_variables_size;  /* how many bytes are in use */

  uint64_t myisam_max_extra_sort_file_size;
  uint64_t max_heap_table_size;
  uint64_t tmp_table_size;
  ha_rows select_limit;
  ha_rows max_join_size;
  uint64_t auto_increment_increment;
  uint64_t auto_increment_offset;
  uint64_t bulk_insert_buff_size;
  uint64_t join_buff_size;
  uint32_t max_allowed_packet;
  uint64_t max_error_count;
  uint64_t max_length_for_sort_data;
  size_t max_sort_length;
  uint64_t min_examined_row_limit;
  bool optimizer_prune_level;
  bool log_warnings;

  uint32_t optimizer_search_depth;
  uint32_t div_precincrement;
  uint64_t preload_buff_size;
  uint32_t read_buff_size;
  uint32_t read_rnd_buff_size;
  bool replicate_query;
  size_t sortbuff_size;
  uint32_t thread_handling;
  uint32_t tx_isolation;
  size_t transaction_message_threshold;
  uint32_t completion_type;
  /* Determines which non-standard SQL behaviour should be enabled */
  uint32_t sql_mode;
  uint64_t max_seeks_for_key;
  size_t range_alloc_block_size;
  uint32_t query_alloc_block_size;
  uint32_t query_prealloc_size;
  uint64_t group_concat_max_len;
  uint64_t pseudo_thread_id;

  plugin::StorageEngine *storage_engine;

  /* Only charset part of these variables is sensible */
  const CHARSET_INFO  *character_set_filesystem;

  /* Both charset and collation parts of these variables are important */
  const CHARSET_INFO	*collation_server;

  inline const CHARSET_INFO  *getCollation(void) 
  {
    return collation_server;
  }

  /* Locale Support */
  MY_LOCALE *lc_time_names;

  Time_zone *time_zone;
};

extern DRIZZLED_API struct drizzle_system_variables global_system_variables;

} /* namespace drizzled */

#include "drizzled/sql_lex.h"

namespace drizzled
{

DRIZZLED_API void mark_transaction_to_rollback(Session *session, bool all);

/**
  Storage engine specific thread local data.
*/
struct Ha_data
{
  /**
    Storage engine specific thread local data.
    Lifetime: one user connection.
  */
  void *ha_ptr;
  /**
   * Resource contexts for both the "statement" and "normal"
   * transactions.
   *
   * Resource context at index 0:
   *
   * Life time: one statement within a transaction. If @@autocommit is
   * on, also represents the entire transaction.
   *
   * Resource context at index 1:
   *
   * Life time: one transaction within a connection. 
   *
   * @note
   *
   * If the storage engine does not participate in a transaction, 
   * there will not be a resource context.
   */
  drizzled::ResourceContext resource_context[2];

  Ha_data() :ha_ptr(NULL) {}
};

/**
 * Represents a client connection to the database server.
 *
 * Contains the client/server protocol object, the current statement
 * being executed, local-to-session variables and status counters, and
 * a host of other information.
 *
 * @todo
 *
 * The Session class should have a vector of Statement object pointers which
 * comprise the statements executed on the Session. Until this architectural
 * change is done, we can forget about parallel operations inside a session.
 *
 * @todo
 *
 * Make member variables private and have inlined accessors and setters.  Hide
 * all member variables that are not critical to non-internal operations of the
 * session object.
 */

class DRIZZLED_API Session :
  public Open_tables_state
{
public:
  // Plugin storage in Session.
  typedef boost::unordered_map<std::string, util::Storable *, util::insensitive_hash, util::insensitive_equal_to> PropertyMap;
  typedef Session* Ptr;
  typedef boost::shared_ptr<Session> shared_ptr;
  typedef Session& reference;
  typedef const Session& const_reference;
  typedef const Session* const_pointer;
  typedef Session* pointer;

  static shared_ptr make_shared(plugin::Client *client, catalog::Instance::shared_ptr instance_arg)
  {
    assert(instance_arg);
    return boost::make_shared<Session>(client, instance_arg);
  }

  /*
    MARK_COLUMNS_NONE:  Means mark_used_colums is not set and no indicator to
                        handler of fields used is set
    MARK_COLUMNS_READ:  Means a bit in read set is set to inform handler
	                that the field is to be read. If field list contains
                        duplicates, then session->dup_field is set to point
                        to the last found duplicate.
    MARK_COLUMNS_WRITE: Means a bit is set in write set to inform handler
			that it needs to update this field in write_row
                        and update_row.
  */
  enum enum_mark_columns mark_used_columns;
  inline void* alloc(size_t size)
  {
    return mem_root->alloc_root(size);
  }
  inline void* calloc(size_t size)
  {
    void *ptr;
    if ((ptr= mem_root->alloc_root(size)))
      memset(ptr, 0, size);
    return ptr;
  }
  inline char *strdup(const char *str)
  {
    return mem_root->strdup_root(str);
  }
  inline char *strmake(const char *str, size_t size)
  {
    return mem_root->strmake_root(str,size);
  }
  inline void *memdup(const void *str, size_t size)
  {
    return mem_root->memdup_root(str, size);
  }
  inline void *memdup_w_gap(const void *str, size_t size, uint32_t gap)
  {
    void *ptr;
    if ((ptr= mem_root->alloc_root(size + gap)))
      memcpy(ptr,str,size);
    return ptr;
  }
  /** Frees all items attached to this Statement */
  void free_items();
  /**
   * List of items created in the parser for this query. Every item puts
   * itself to the list on creation (see Item::Item() for details))
   */
  Item *free_list;
  memory::Root *mem_root; /**< Pointer to current memroot */


  memory::Root *getMemRoot()
  {
    return mem_root;
  }

  uint64_t xa_id;

  uint64_t getXaId()
  {
    return xa_id;
  }

  void setXaId(uint64_t in_xa_id)
  {
    xa_id= in_xa_id; 
  }

  /**
   * Uniquely identifies each statement object in thread scope; change during
   * statement lifetime.
   *
   * @todo should be const
   */
  uint32_t id;
  LEX *lex; /**< parse tree descriptor */

  LEX *getLex() 
  {
    return lex;
  }

  enum_sql_command getSqlCommand() const
  {
    return lex->sql_command;
  }

  /** query associated with this statement */
  typedef boost::shared_ptr<const std::string> QueryString;
private:
  boost::shared_ptr<std::string> query;

  // Never allow for a modification of this outside of the class. c_str()
  // requires under some setup non const, you must copy the QueryString in
  // order to use it.
public:
  QueryString getQueryString() const
  {
    return query;
  }

  void resetQueryString()
  {
    query.reset();
    _state.reset();
  }

  /*
    We need to copy the lock on the string in order to make sure we have a stable string.
    Once this is done we can use it to build a const char* which can be handed off for
    a method to use (Innodb is currently the only engine using this).
  */
  const char *getQueryStringCopy(size_t &length)
  {
    QueryString tmp_string(getQueryString());

    if (not tmp_string)
    {
      length= 0;
      return NULL;
    }

    length= tmp_string->length();
    char *to_return= strmake(tmp_string->c_str(), tmp_string->length());
    return to_return;
  }

  class State {
    std::vector <char> _query;

  public:
    typedef boost::shared_ptr<State> const_shared_ptr;

    State(const char *in_packet, size_t in_packet_length)
    {
      if (in_packet_length)
      {
        size_t minimum= std::min(in_packet_length, static_cast<size_t>(PROCESS_LIST_WIDTH));
        _query.resize(minimum + 1);
        memcpy(&_query[0], in_packet, minimum);
      }
      else
      {
        _query.resize(0);
      }
    }

    const char *query() const
    {
      if (_query.size())
        return &_query[0];

      return "";
    }

    const char *query(size_t &size) const
    {
      if (_query.size())
      {
        size= _query.size() -1;
        return &_query[0];
      }

      size= 0;
      return "";
    }
  protected:
    friend class Session;
    typedef boost::shared_ptr<State> shared_ptr;
  };
private:
  State::shared_ptr  _state; 
public:

  State::const_shared_ptr state()
  {
    return _state;
  }

  /**
    Name of the current (default) database.

    If there is the current (default) database, "db" contains its name. If
    there is no current (default) database, "db" is NULL and "db_length" is
    0. In other words, "db", "db_length" must either be NULL, or contain a
    valid database name.

    @note this attribute is set and alloced by the slave SQL thread (for
    the Session of that thread); that thread is (and must remain, for now) the
    only responsible for freeing this member.
  */
private:
  util::string::shared_ptr _schema;
public:

  util::string::const_shared_ptr schema() const
  {
    if (_schema)
      return _schema;

    return util::string::const_shared_ptr(new std::string(""));
  }

  /* current cache key */
  std::string query_cache_key;
  /**
    Constant for Session::where initialization in the beginning of every query.

    It's needed because we do not save/restore Session::where normally during
    primary (non subselect) query execution.
  */
  static const char * const DEFAULT_WHERE;

  memory::Root warn_root; /**< Allocation area for warnings and errors */
private:
  plugin::Client *client; /**< Pointer to client object */

public:

  void setClient(plugin::Client *client_arg);

  plugin::Client *getClient()
  {
    return client;
  }

  plugin::Client *getClient() const
  {
    return client;
  }

  plugin::Scheduler *scheduler; /**< Pointer to scheduler object */
  void *scheduler_arg; /**< Pointer to the optional scheduler argument */

  typedef boost::unordered_map< std::string, user_var_entry *, util::insensitive_hash, util::insensitive_equal_to> UserVars;
private:
  typedef std::pair< UserVars::iterator, UserVars::iterator > UserVarsRange;
  UserVars user_vars; /**< Hash of user variables defined during the session's lifetime */

public:

  const UserVars &getUserVariables() const
  {
    return user_vars;
  }

  drizzle_system_variables variables; /**< Mutable local variables local to the session */

  enum_tx_isolation getTxIsolation()
  {
    return (enum_tx_isolation)variables.tx_isolation;
  }

  struct system_status_var status_var; /**< Session-local status counters */
  THR_LOCK_INFO lock_info; /**< Locking information for this session */
  THR_LOCK_OWNER main_lock_id; /**< To use for conventional queries */
  THR_LOCK_OWNER *lock_id; /**< If not main_lock_id, points to the lock_id of a cursor. */

  /**
   * A pointer to the stack frame of the scheduler thread
   * which is called first in the thread for handling a client
   */
  char *thread_stack;

private:
  identifier::User::shared_ptr security_ctx;

  int32_t scoreboard_index;

  inline void checkSentry() const
  {
    assert(this->dbug_sentry == Session_SENTRY_MAGIC);
  }
public:
  identifier::User::const_shared_ptr user() const
  {
    if (security_ctx)
      return security_ctx;

    return identifier::User::const_shared_ptr();
  }

  void setUser(identifier::User::shared_ptr arg)
  {
    security_ctx= arg;
  }

  int32_t getScoreboardIndex()
  {
    return scoreboard_index;
  }

  void setScoreboardIndex(int32_t in_scoreboard_index)
  {
    scoreboard_index= in_scoreboard_index;
  }

  /**
   * Is this session viewable by the current user?
   */
  bool isViewable(identifier::User::const_reference) const;

private:
  /**
    Used in error messages to tell user in what part of MySQL we found an
    error. E. g. when where= "having clause", if fix_fields() fails, user
    will know that the error was in having clause.
  */
  const char *_where;

public:
  const char *where()
  {
    return _where;
  }

  void setWhere(const char *arg)
  {
    _where= arg;
  }

  /*
    One thread can hold up to one named user-level lock. This variable
    points to a lock object if the lock is present. See item_func.cc and
    chapter 'Miscellaneous functions', for functions GET_LOCK, RELEASE_LOCK.
  */
  uint32_t dbug_sentry; /**< watch for memory corruption */

private:
  boost::thread::id boost_thread_id;
  boost_thread_shared_ptr _thread;
  boost::this_thread::disable_interruption *interrupt;

  internal::st_my_thread_var *mysys_var;
public:

  boost_thread_shared_ptr &getThread()
  {
    return _thread;
  }

  void pushInterrupt(boost::this_thread::disable_interruption *interrupt_arg)
  {
    interrupt= interrupt_arg;
  }

  boost::this_thread::disable_interruption &getThreadInterupt()
  {
    assert(interrupt);
    return *interrupt;
  }

  internal::st_my_thread_var *getThreadVar()
  {
    return mysys_var;
  }

  /**
   * Type of current query: COM_STMT_PREPARE, COM_QUERY, etc. Set from
   * first byte of the packet in executeStatement()
   */
  enum enum_server_command command;
  uint32_t file_id;	/**< File ID for LOAD DATA INFILE */
  /* @note the following three members should likely move to Client */
  uint32_t max_client_packet_length; /**< Maximum number of bytes a client can send in a single packet */

private:
  boost::posix_time::ptime _epoch;
  boost::posix_time::ptime _connect_time;
  boost::posix_time::ptime _start_timer;
  boost::posix_time::ptime _end_timer;

  boost::posix_time::ptime _user_time;
public:
  uint64_t utime_after_lock; // This used by Innodb.

  void resetUserTime()
  {
    _user_time= boost::posix_time::not_a_date_time;
  }

  const boost::posix_time::ptime &start_timer() const
  {
    return _start_timer;
  }

  void getTimeDifference(boost::posix_time::time_duration &result_arg, const boost::posix_time::ptime &arg) const
  {
    result_arg=  arg - _start_timer;
  }

  thr_lock_type update_lock_default;

  /*
    Both of the following container points in session will be converted to an API.
  */

private:
  /* container for handler's private per-connection data */
  std::vector<Ha_data> ha_data;
  /*
    Id of current query. Statement can be reused to execute several queries
    query_id is global in context of the whole MySQL server.
    ID is automatically generated from an atomic counter.
    It's used in Cursor code for various purposes: to check which columns
    from table are necessary for this select, to check if it's necessary to
    update auto-updatable fields (like auto_increment and timestamp).
  */
  query_id_t query_id;
  query_id_t warn_query_id;

public:
  void **getEngineData(const plugin::MonitoredInTransaction *monitored);
  ResourceContext *getResourceContext(const plugin::MonitoredInTransaction *monitored,
                                      size_t index= 0);

  /**
   * Structure used to manage "statement transactions" and
   * "normal transactions". In autocommit mode, the normal transaction is
   * equivalent to the statement transaction.
   *
   * Storage engines will be registered here when they participate in
   * a transaction. No engine is registered more than once.
   */
  struct st_transactions {
    std::deque<NamedSavepoint> savepoints;

    /**
     * The normal transaction (since BEGIN WORK).
     *
     * Contains a list of all engines that have participated in any of the
     * statement transactions started within the context of the normal
     * transaction.
     *
     * @note In autocommit mode, this is empty.
     */
    TransactionContext all;

    /**
     * The statment transaction.
     *
     * Contains a list of all engines participating in the given statement.
     *
     * @note In autocommit mode, this will be used to commit/rollback the
     * normal transaction.
     */
    TransactionContext stmt;

    XID_STATE xid_state;

    void cleanup()
    {
      savepoints.clear();
    }
    st_transactions() :
      savepoints(),
      all(),
      stmt(),
      xid_state()
    { }
  } transaction;

  Field *dup_field;
  sigset_t signals;

  // As of right now we do not allow a concurrent execute to launch itself
private:
  bool concurrent_execute_allowed;
public:

  void setConcurrentExecute(bool arg)
  {
    concurrent_execute_allowed= arg;
  }

  bool isConcurrentExecuteAllowed() const
  {
    return concurrent_execute_allowed;
  }

  /* Tells if LAST_INSERT_ID(#) was called for the current statement */
  bool arg_of_last_insert_id_function;

  /*
    ALL OVER THIS FILE, "insert_id" means "*automatically generated* value for
    insertion into an auto_increment column".
  */
  /**
    This is the first autogenerated insert id which was *successfully*
    inserted by the previous statement (exactly, if the previous statement
    didn't successfully insert an autogenerated insert id, then it's the one
    of the statement before, etc).
    It can also be set by SET LAST_INSERT_ID=# or SELECT LAST_INSERT_ID(#).
    It is returned by LAST_INSERT_ID().
  */
  uint64_t first_successful_insert_id_in_prev_stmt;
  /**
    This is the first autogenerated insert id which was *successfully*
    inserted by the current statement. It is maintained only to set
    first_successful_insert_id_in_prev_stmt when statement ends.
  */
  uint64_t first_successful_insert_id_in_cur_stmt;
  /**
    We follow this logic:
    - when stmt starts, first_successful_insert_id_in_prev_stmt contains the
    first insert id successfully inserted by the previous stmt.
    - as stmt makes progress, handler::insert_id_for_cur_row changes;
    every time get_auto_increment() is called,
    auto_inc_intervals_in_cur_stmt_for_binlog is augmented with the
    reserved interval (if statement-based binlogging).
    - at first successful insertion of an autogenerated value,
    first_successful_insert_id_in_cur_stmt is set to
    handler::insert_id_for_cur_row.
    - when stmt goes to binlog,
    auto_inc_intervals_in_cur_stmt_for_binlog is binlogged if
    non-empty.
    - when stmt ends, first_successful_insert_id_in_prev_stmt is set to
    first_successful_insert_id_in_cur_stmt.

    List of auto_increment intervals reserved by the thread so far, for
    storage in the statement-based binlog.
    Note that its minimum is not first_successful_insert_id_in_cur_stmt:
    assuming a table with an autoinc column, and this happens:
    INSERT INTO ... VALUES(3);
    SET INSERT_ID=3; INSERT IGNORE ... VALUES (NULL);
    then the latter INSERT will insert no rows
    (first_successful_insert_id_in_cur_stmt == 0), but storing "INSERT_ID=3"
    in the binlog is still needed; the list's minimum will contain 3.
  */
  Discrete_intervals_list auto_inc_intervals_in_cur_stmt_for_binlog;
  /** Used by replication and SET INSERT_ID */
  Discrete_intervals_list auto_inc_intervals_forced;

  uint64_t limit_found_rows;
  uint64_t options; /**< Bitmap of options */
  int64_t row_count_func; /**< For the ROW_COUNT() function */

  int64_t rowCount() const
  {
    return row_count_func;
  }

  ha_rows cuted_fields; /**< Count of "cut" or truncated fields. @todo Kill this friggin thing. */

  /** 
   * Number of rows we actually sent to the client, including "synthetic"
   * rows in ROLLUP etc.
   */
  ha_rows sent_row_count;

  /**
   * Number of rows we read, sent or not, including in create_sort_index()
   */
  ha_rows examined_row_count;

  /**
   * The set of those tables whose fields are referenced in all subqueries
   * of the query.
   *
   * @todo
   * 
   * Possibly this it is incorrect to have used tables in Session because
   * with more than one subquery, it is not clear what does the field mean.
   */
  table_map used_tables;

  /**
    @todo
    
    This, and some other variables like 'count_cuted_fields'
    maybe should be statement/cursor local, that is, moved to Statement
    class. With current implementation warnings produced in each prepared
    statement/cursor settle here.
  */
  List<DRIZZLE_ERROR> warn_list;
  uint32_t warn_count[(uint32_t) DRIZZLE_ERROR::WARN_LEVEL_END];
  uint32_t total_warn_count;
  Diagnostics_area main_da;

  ulong col_access;

  /* Statement id is thread-wide. This counter is used to generate ids */
  uint32_t statement_id_counter;
  uint32_t rand_saved_seed1;
  uint32_t rand_saved_seed2;
  /**
    Row counter, mainly for errors and warnings. Not increased in
    create_sort_index(); may differ from examined_row_count.
  */
  uint32_t row_count;

  uint32_t getRowCount() const
  {
    return row_count;
  }

  session_id_t thread_id;
  uint32_t tmp_table;
  enum global_read_lock_t
  {
    NONE= 0,
    GOT_GLOBAL_READ_LOCK= 1,
    MADE_GLOBAL_READ_LOCK_BLOCK_COMMIT= 2
  };
private:
  global_read_lock_t _global_read_lock;

public:

  global_read_lock_t isGlobalReadLock() const
  {
    return _global_read_lock;
  }

  void setGlobalReadLock(global_read_lock_t arg)
  {
    _global_read_lock= arg;
  }

  DrizzleLock *lockTables(Table **tables, uint32_t count, uint32_t flags);
  bool lockGlobalReadLock();
  bool lock_table_names(TableList *table_list);
  bool lock_table_names_exclusively(TableList *table_list);
  bool makeGlobalReadLockBlockCommit();
  bool abortLockForThread(Table *table);
  bool wait_if_global_read_lock(bool abort_on_refresh, bool is_not_commit);
  int lock_table_name(TableList *table_list);
  void abortLock(Table *table);
  void removeLock(Table *table);
  void unlockReadTables(DrizzleLock *sql_lock);
  void unlockSomeTables(Table **table, uint32_t count);
  void unlockTables(DrizzleLock *sql_lock);
  void startWaitingGlobalReadLock();
  void unlockGlobalReadLock();

private:
  int unlock_external(Table **table, uint32_t count);
  int lock_external(Table **tables, uint32_t count);
  bool wait_for_locked_table_names(TableList *table_list);
  DrizzleLock *get_lock_data(Table **table_ptr, uint32_t count,
                             bool should_lock, Table **write_lock_used);
public:

  uint32_t server_status;
  uint32_t open_options;
  uint32_t select_number; /**< number of select (used for EXPLAIN) */
  /* variables.transaction_isolation is reset to this after each commit */
  enum_tx_isolation session_tx_isolation;
  enum_check_fields count_cuted_fields;

  enum killed_state_t
  {
    NOT_KILLED,
    KILL_BAD_DATA,
    KILL_CONNECTION,
    KILL_QUERY,
    KILLED_NO_VALUE /* means none of the above states apply */
  };
private:
  killed_state_t volatile _killed;

public:

  void setKilled(killed_state_t arg)
  {
    _killed= arg;
  }

  killed_state_t getKilled()
  {
    return _killed;
  }

  volatile killed_state_t *getKilledPtr() // Do not use this method, it is here for historical convience.
  {
    return &_killed;
  }

  bool is_admin_connection;
  bool some_tables_deleted;
  bool no_errors;
  bool password;
  /**
    Set to true if execution of the current compound statement
    can not continue. In particular, disables activation of
    CONTINUE or EXIT handlers of stored routines.
    Reset in the end of processing of the current user request, in
    @see reset_session_for_next_command().
  */
  bool is_fatal_error;
  /**
    Set by a storage engine to request the entire
    transaction (that possibly spans multiple engines) to
    rollback. Reset in ha_rollback.
  */
  bool transaction_rollback_request;
  /**
    true if we are in a sub-statement and the current error can
    not be safely recovered until we left the sub-statement mode.
    In particular, disables activation of CONTINUE and EXIT
    handlers inside sub-statements. E.g. if it is a deadlock
    error and requires a transaction-wide rollback, this flag is
    raised (traditionally, MySQL first has to close all the reads
    via @see handler::ha_index_or_rnd_end() and only then perform
    the rollback).
    Reset to false when we leave the sub-statement mode.
  */
  bool is_fatal_sub_stmt_error;
  /** for IS NULL => = last_insert_id() fix in remove_eq_conds() */
  bool substitute_null_with_insert_id;
  bool cleanup_done;

private:
  bool abort_on_warning;
  bool tablespace_op; /**< This is true in DISCARD/IMPORT TABLESPACE */

public:
  bool got_warning; /**< Set on call to push_warning() */
  bool no_warnings_for_error; /**< no warnings on call to my_error() */
  /** set during loop of derived table processing */
  bool derived_tables_processing;

  bool doing_tablespace_operation(void)
  {
    return tablespace_op;
  }

  void setDoingTablespaceOperation(bool doing)
  {
    tablespace_op= doing;
  }


  /** Used by the sys_var class to store temporary values */
  union
  {
    bool bool_value;
    uint32_t uint32_t_value;
    int32_t int32_t_value;
    uint64_t uint64_t_value;
  } sys_var_tmp;

  /**
    Character input stream consumed by the lexical analyser,
    used during parsing.
    Note that since the parser is not re-entrant, we keep only one input
    stream here. This member is valid only when executing code during parsing,
    and may point to invalid memory after that.
  */
  Lex_input_stream *m_lip;
  
  /** Place to store various things */
  void *session_marker;

  /** Keeps a copy of the previous table around in case we are just slamming on particular table */
  Table *cached_table;

  /**
    Points to info-string that we show in SHOW PROCESSLIST
    You are supposed to call Session_SET_PROC_INFO only if you have coded
    a time-consuming piece that MySQL can get stuck in for a long time.

    Set it using the  session_proc_info(Session *thread, const char *message)
    macro/function.
  */
  inline void set_proc_info(const char *info)
  { 
    proc_info= info;
  }
  inline const char* get_proc_info() const
  {
    return proc_info;
  }

  /** Sets this Session's current query ID */
  inline void setQueryId(query_id_t in_query_id)
  {
    query_id= in_query_id;
  }

  /** Returns the current query ID */
  query_id_t getQueryId()  const
  {
    return query_id;
  }


  /** Sets this Session's warning query ID */
  inline void setWarningQueryId(query_id_t in_query_id)
  {
    warn_query_id= in_query_id;
  }

  /** Returns the Session's warning query ID */
  inline query_id_t getWarningQueryId()  const
  {
    return warn_query_id;
  }

  /** Accessor method returning the session's ID. */
  inline session_id_t getSessionId()  const
  {
    return thread_id;
  }

  /** Accessor method returning the server's ID. */
  inline uint32_t getServerId()  const
  {
    /* We return the global server ID. */
    return server_id;
  }

  /** Returns the current transaction ID for the session's current statement */
  inline my_xid getTransactionId()
  {
    return transaction.xid_state.xid.quick_get_my_xid();
  }
  /**
    There is BUG#19630 where statement-based replication of stored
    functions/triggers with two auto_increment columns breaks.
    We however ensure that it works when there is 0 or 1 auto_increment
    column; our rules are
    a) on master, while executing a top statement involving substatements,
    first top- or sub- statement to generate auto_increment values wins the
    exclusive right to see its values be written to binlog (the write
    will be done by the statement or its caller), and the losers won't see
    their values be written to binlog.
    b) on slave, while replicating a top statement involving substatements,
    first top- or sub- statement to need to read auto_increment values from
    the master's binlog wins the exclusive right to read them (so the losers
    won't read their values from binlog but instead generate on their own).
    a) implies that we mustn't backup/restore
    auto_inc_intervals_in_cur_stmt_for_binlog.
    b) implies that we mustn't backup/restore auto_inc_intervals_forced.

    If there are more than 1 auto_increment columns, then intervals for
    different columns may mix into the
    auto_inc_intervals_in_cur_stmt_for_binlog list, which is logically wrong,
    but there is no point in preventing this mixing by preventing intervals
    from the secondly inserted column to come into the list, as such
    prevention would be wrong too.
    What will happen in the case of
    INSERT INTO t1 (auto_inc) VALUES(NULL);
    where t1 has a trigger which inserts into an auto_inc column of t2, is
    that in binlog we'll store the interval of t1 and the interval of t2 (when
    we store intervals, soon), then in slave, t1 will use both intervals, t2
    will use none; if t1 inserts the same number of rows as on master,
    normally the 2nd interval will not be used by t1, which is fine. t2's
    values will be wrong if t2's internal auto_increment counter is different
    from what it was on master (which is likely). In 5.1, in mixed binlogging
    mode, row-based binlogging is used for such cases where two
    auto_increment columns are inserted.
  */
  inline void record_first_successful_insert_id_in_cur_stmt(uint64_t id_arg)
  {
    if (first_successful_insert_id_in_cur_stmt == 0)
      first_successful_insert_id_in_cur_stmt= id_arg;
  }
  inline uint64_t read_first_successful_insert_id_in_prev_stmt(void)
  {
    return first_successful_insert_id_in_prev_stmt;
  }
  /**
    Used by Intvar_log_event::do_apply_event() and by "SET INSERT_ID=#"
    (mysqlbinlog). We'll soon add a variant which can take many intervals in
    argument.
  */
  inline void force_one_auto_inc_interval(uint64_t next_id)
  {
    auto_inc_intervals_forced.empty(); // in case of multiple SET INSERT_ID
    auto_inc_intervals_forced.append(next_id, UINT64_MAX, 0);
  }

  Session(plugin::Client *client_arg, catalog::Instance::shared_ptr catalog);
  virtual ~Session();

  void cleanup(void);
  /**
   * Cleans up after query.
   *
   * @details
   *
   * This function is used to reset thread data to its default state.
   *
   * This function is not suitable for setting thread data to some
   * non-default values, as there is only one replication thread, so
   * different master threads may overwrite data of each other on
   * slave.
   */
  void cleanup_after_query();
  bool storeGlobals();
  void awake(Session::killed_state_t state_to_set);
  /**
   * Pulls thread-specific variables into Session state.
   *
   * Returns true most times, or false if there was a problem
   * allocating resources for thread-specific storage.
   *
   * @TODO Kill this.  It's not necessary once my_thr_init() is bye bye.
   *
   */
  bool initGlobals();

  /**
    Initialize memory roots necessary for query processing and (!)
    pre-allocate memory for it. We can't do that in Session constructor because
    there are use cases where it's vital to not allocate excessive and not used
    memory.
  */
  void prepareForQueries();

  /**
   * Executes a single statement received from the 
   * client connection.
   *
   * Returns true if the statement was successful, or false 
   * otherwise.
   *
   * @note
   *
   * For profiling to work, it must never be called recursively.
   *
   * In MySQL, this used to be the do_command() C function whic
   * accepted a single parameter of the THD pointer.
   */
  bool executeStatement();

  /**
   * Reads a query from packet and stores it.
   *
   * Returns true if query is read and allocated successfully, 
   * false otherwise.  On a return of false, Session::fatal_error
   * is set.
   *
   * @note Used in COM_QUERY and COM_STMT_PREPARE.
   *
   * Sets the following Session variables:
   *  - query
   *  - query_length
   *
   * @param The packet pointer to read from
   * @param The length of the query to read
   */
  bool readAndStoreQuery(const char *in_packet, uint32_t in_packet_length);

  /**
   * Ends the current transaction and (maybe) begins the next.
   *
   * Returns true if the transaction completed successfully, 
   * otherwise false.
   *
   * @param Completion type
   */
  bool endTransaction(enum enum_mysql_completiontype completion);
  bool endActiveTransaction();
  bool startTransaction(start_transaction_option_t opt= START_TRANS_NO_OPTIONS);

  /**
   * Authenticates users, with error reporting.
   *
   * Returns true on success, or false on failure.
   */
  bool authenticate();

  /**
   * Run a session.
   *
   * This will initialize the session and begin the command loop.
   */
  void run();

  /**
   * Schedule a session to be run on the default scheduler.
   */
  static bool schedule(Session::shared_ptr&);

  static void unlink(session_id_t &session_id);
  static void unlink(Session::shared_ptr&);

  /*
    For enter_cond() / exit_cond() to work the mutex must be got before
    enter_cond(); this mutex is then released by exit_cond().
    Usage must be: lock mutex; enter_cond(); your code; exit_cond().
  */
  const char* enter_cond(boost::condition_variable_any &cond, boost::mutex &mutex, const char* msg);
  void exit_cond(const char* old_msg);

  type::Time::epoch_t query_start()
  {
    return getCurrentTimestampEpoch();
  }

  void set_time()
  {
    _end_timer= _start_timer= boost::posix_time::microsec_clock::universal_time();
    utime_after_lock= (_start_timer - _epoch).total_microseconds();
  }

  void set_time(time_t t) // This is done by a sys_var, as long as user_time is set, we will use that for all references to time
  {
    _user_time= boost::posix_time::from_time_t(t);
  }

  void set_time_after_lock()
  { 
    boost::posix_time::ptime mytime(boost::posix_time::microsec_clock::universal_time());
    utime_after_lock= (mytime - _epoch).total_microseconds();
  }

  void set_end_timer()
  {
    _end_timer= boost::posix_time::microsec_clock::universal_time();
    status_var.execution_time_nsec+=(_end_timer - _start_timer).total_microseconds();
  }

  uint64_t getElapsedTime() const
  {
    return (_end_timer - _start_timer).total_microseconds();
  }

  /**
   * Returns the current micro-timestamp
   */
  type::Time::epoch_t getCurrentTimestamp(bool actual= true) const
  { 
    type::Time::epoch_t t_mark;

    if (actual)
    {
      boost::posix_time::ptime mytime(boost::posix_time::microsec_clock::universal_time());
      t_mark= (mytime - _epoch).total_microseconds();
    }
    else
    {
      t_mark= (_end_timer - _epoch).total_microseconds();
    }

    return t_mark; 
  }

  // We may need to set user on this
  type::Time::epoch_t getCurrentTimestampEpoch() const
  { 
    if (not _user_time.is_not_a_date_time())
      return (_user_time - _epoch).total_seconds();

    return (_start_timer - _epoch).total_seconds();
  }

  type::Time::epoch_t getCurrentTimestampEpoch(type::Time::usec_t &fraction_arg) const
  { 
    if (not _user_time.is_not_a_date_time())
    {
      fraction_arg= 0;
      return (_user_time - _epoch).total_seconds();
    }

    fraction_arg= _start_timer.time_of_day().fractional_seconds() % 1000000;
    return (_start_timer - _epoch).total_seconds();
  }

  uint64_t found_rows(void) const
  {
    return limit_found_rows;
  }

  /** Returns whether the session is currently inside a transaction */
  bool inTransaction() const
  {
    return server_status & SERVER_STATUS_IN_TRANS;
  }

  LEX_STRING *make_lex_string(LEX_STRING *lex_str,
                              const char* str, uint32_t length,
                              bool allocate_lex_string);

  LEX_STRING *make_lex_string(LEX_STRING *lex_str,
                              const std::string &str,
                              bool allocate_lex_string);

  int send_explain_fields(select_result *result);

  /**
    Clear the current error, if any.
    We do not clear is_fatal_error or is_fatal_sub_stmt_error since we
    assume this is never called if the fatal error is set.
    @todo: To silence an error, one should use Internal_error_handler
    mechanism. In future this function will be removed.
  */
  inline void clear_error(bool full= false)
  {
    if (main_da.is_error())
      main_da.reset_diagnostics_area();

    if (full)
    {
      drizzle_reset_errors(this, true);
    }
  }

  void clearDiagnostics()
  {
    main_da.reset_diagnostics_area();
  }

  /**
    Mark the current error as fatal. Warning: this does not
    set any error, it sets a property of the error, so must be
    followed or prefixed with my_error().
  */
  inline void fatal_error()
  {
    assert(main_da.is_error());
    is_fatal_error= true;
  }
  /**
    true if there is an error in the error stack.

    Please use this method instead of direct access to
    net.report_error.

    If true, the current (sub)-statement should be aborted.
    The main difference between this member and is_fatal_error
    is that a fatal error can not be handled by a stored
    procedure continue handler, whereas a normal error can.

    To raise this flag, use my_error().
  */
  inline bool is_error() const { return main_da.is_error(); }
  inline const CHARSET_INFO *charset() { return default_charset_info; }

  void change_item_tree(Item **place, Item *new_value)
  {
    *place= new_value;
  }
  /**
    Cleanup statement parse state (parse tree, lex) and execution
    state after execution of a non-prepared SQL statement.

    @todo

    Move this to Statement::~Statement
  */
  void end_statement();
  inline int killed_errno() const
  {
    killed_state_t killed_val; /* to cache the volatile 'killed' */
    return (killed_val= _killed) != KILL_BAD_DATA ? killed_val : 0;
  }
  void send_kill_message() const;
  /* return true if we will abort query if we make a warning now */
  inline bool abortOnWarning()
  {
    return abort_on_warning;
  }

  inline void setAbortOnWarning(bool arg)
  {
    abort_on_warning= arg;
  }

  void setAbort(bool arg);
  void lockOnSys();
  void set_status_var_init();

  /**
    Set the current database; use deep copy of C-string.

    @param new_db     a pointer to the new database name.
    @param new_db_len length of the new database name.

    Initialize the current database from a NULL-terminated string with
    length. If we run out of memory, we free the current database and
    return true.  This way the user will notice the error as there will be
    no current database selected (in addition to the error message set by
    malloc).

    @note This operation just sets {db, db_length}. Switching the current
    database usually involves other actions, like switching other database
    attributes including security context. In the future, this operation
    will be made private and more convenient interface will be provided.
  */
  void set_db(const std::string &new_db);

  /*
    Copy the current database to the argument. Use the current arena to
    allocate memory for a deep copy: current database may be freed after
    a statement is parsed but before it's executed.
  */
  bool copy_db_to(char **p_db, size_t *p_db_length);

public:
  /**
    Add an internal error handler to the thread execution context.
    @param handler the exception handler to add
  */
  void push_internal_handler(Internal_error_handler *handler);

  /**
    Handle an error condition.
    @param sql_errno the error number
    @param level the error level
    @return true if the error is handled
  */
  virtual bool handle_error(drizzled::error_t sql_errno, const char *message,
                            DRIZZLE_ERROR::enum_warning_level level);

  /**
    Remove the error handler last pushed.
  */
  void pop_internal_handler();

  /**
    Resets Session part responsible for command processing state.

    This needs to be called before execution of every statement
    (prepared or conventional).
    It is not called by substatements of routines.

    @todo
    Make it a method of Session and align its name with the rest of
    reset/end/start/init methods.
    @todo
    Call it after we use Session for queries, not before.
  */
  void reset_for_next_command();

  /**
   * Disconnects the session from a client connection and
   * updates any status variables necessary.
   *
   * @param errcode	Error code to print to console
   *
   * @note  For the connection that is doing shutdown, this is called twice
   */
  void disconnect(enum error_t errcode= EE_OK);

  /**
   * Check if user exists and the password supplied is correct.
   *
   * Returns true on success, and false on failure.
   *
   * @note Host, user and passwd may point to communication buffer.
   * Current implementation does not depend on that, but future changes
   * should be done with this in mind; 
   *
   * @param passwd Scrambled password received from client
   * @param db Database name to connect to, may be NULL
   */
  bool checkUser(const std::string &passwd, const std::string &db);
  
  /**
   * Returns the timestamp (in microseconds) of when the Session 
   * connected to the server.
   */
  uint64_t getConnectMicroseconds() const
  {
    return (_connect_time - _epoch).total_microseconds();
  }

  uint64_t getConnectSeconds() const
  {
    return (_connect_time - _epoch).total_seconds();
  }

  /**
   * Returns a pointer to the active Transaction message for this
   * Session being managed by the ReplicationServices component, or
   * NULL if no active message.
   */
  message::Transaction *getTransactionMessage() const
  {
    return transaction_message;
  }

  /**
   * Returns a pointer to the active Statement message for this
   * Session, or NULL if no active message.
   */
  message::Statement *getStatementMessage() const
  {
    return statement_message;
  }
  
  /**
   * Returns a pointer to the current Resulset message for this
   * Session, or NULL if no active message.
   */
  message::Resultset *getResultsetMessage() const
  {
    return resultset;
  }
  /**
   * Sets the active transaction message used by the ReplicationServices
   * component.
   *
   * @param[in] Pointer to the message
   */
  void setTransactionMessage(message::Transaction *in_message)
  {
    transaction_message= in_message;
  }

  /**
   * Sets the active statement message used by the ReplicationServices
   * component.
   *
   * @param[in] Pointer to the message
   */
  void setStatementMessage(message::Statement *in_message)
  {
    statement_message= in_message;
  }

  /**
   * Sets the active Resultset message used by the Query Cache
   * plugin.
   *
   * @param[in] Pointer to the message
   */
  void setResultsetMessage(message::Resultset *in_message)
  {
    resultset= in_message;
  }
  /**
   * reset the active Resultset message used by the Query Cache
   * plugin.
   */

  void resetResultsetMessage()
  { 
    resultset= NULL;
  }

private:
  /** Pointers to memory managed by the ReplicationServices component */
  message::Transaction *transaction_message;
  message::Statement *statement_message;
  /* Pointer to the current resultset of Select query */
  message::Resultset *resultset;
  plugin::EventObserverList *session_event_observers;
  
  /* Schema observers are mapped to databases. */
  std::map<std::string, plugin::EventObserverList *> schema_event_observers;

 
public:
  plugin::EventObserverList *getSessionObservers() 
  { 
    return session_event_observers;
  }
  
  void setSessionObservers(plugin::EventObserverList *observers) 
  { 
    session_event_observers= observers;
  }
  
  /* For schema event observers there is one set of observers per database. */
  plugin::EventObserverList *getSchemaObservers(const std::string &db_name) 
  { 
    std::map<std::string, plugin::EventObserverList *>::iterator it;
    
    it= schema_event_observers.find(db_name);
    if (it == schema_event_observers.end())
      return NULL;
      
    return it->second;
  }
  
  void setSchemaObservers(const std::string &db_name, plugin::EventObserverList *observers) 
  { 
    std::map<std::string, plugin::EventObserverList *>::iterator it;

    it= schema_event_observers.find(db_name);
    if (it != schema_event_observers.end())
      schema_event_observers.erase(it);;

    if (observers)
      schema_event_observers[db_name] = observers;
  }
  
  
 private:
  const char *proc_info;

  /** The current internal error handler for this thread, or NULL. */
  Internal_error_handler *m_internal_handler;
  /**
    The lex to hold the parsed tree of conventional (non-prepared) queries.
    Whereas for prepared and stored procedure statements we use an own lex
    instance for each new query, for conventional statements we reuse
    the same lex. (@see mysql_parse for details).
  */
  LEX main_lex;
  /**
    This memory root is used for two purposes:
    - for conventional queries, to allocate structures stored in main_lex
    during parsing, and allocate runtime data (execution plan, etc.)
    during execution.
    - for prepared queries, only to allocate runtime data. The parsed
    tree itself is reused between executions and thus is stored elsewhere.
  */
  memory::Root main_mem_root;

  /**
   * Marks all tables in the list which were used by current substatement
   * as free for reuse.
   *
   * @param Head of the list of tables
   *
   * @note
   *
   * The reason we reset query_id is that it's not enough to just test
   * if table->query_id != session->query_id to know if a table is in use.
   *
   * For example
   * 
   *  SELECT f1_that_uses_t1() FROM t1;
   *  
   * In f1_that_uses_t1() we will see one instance of t1 where query_id is
   * set to query_id of original query.
   */
  void mark_used_tables_as_free_for_reuse(Table *table);

public:

  /** A short cut for session->main_da.set_ok_status(). */
  inline void my_ok(ha_rows affected_rows= 0, ha_rows found_rows_arg= 0,
                    uint64_t passed_id= 0, const char *message= NULL)
  {
    main_da.set_ok_status(this, affected_rows, found_rows_arg, passed_id, message);
  }


  /** A short cut for session->main_da.set_eof_status(). */

  inline void my_eof()
  {
    main_da.set_eof_status(this);
  }

  /* Some inline functions for more speed */

  inline bool add_item_to_list(Item *item)
  {
    return lex->current_select->add_item_to_list(this, item);
  }

  inline bool add_value_to_list(Item *value)
  {
    return lex->value_list.push_back(value);
  }

  inline bool add_order_to_list(Item *item, bool asc)
  {
    return lex->current_select->add_order_to_list(this, item, asc);
  }

  inline bool add_group_to_list(Item *item, bool asc)
  {
    return lex->current_select->add_group_to_list(this, item, asc);
  }
  void refresh_status();
  user_var_entry *getVariable(LEX_STRING &name, bool create_if_not_exists);
  user_var_entry *getVariable(const std::string  &name, bool create_if_not_exists);
  void setVariable(const std::string &name, const std::string &value);
  
  /**
   * Closes all tables used by the current substatement, or all tables
   * used by this thread if we are on the upper level.
   */
  void close_thread_tables();
  void close_old_data_files(bool morph_locks= false,
                            bool send_refresh= false);
  void close_open_tables();
  void close_data_files_and_morph_locks(const identifier::Table &identifier);

private:
  bool free_cached_table(boost::mutex::scoped_lock &scopedLock);

public:

  /**
   * Prepares statement for reopening of tables and recalculation of set of
   * prelocked tables.
   *
   * @param Pointer to a pointer to a list of tables which we were trying to open and lock
   */
  void close_tables_for_reopen(TableList **tables);


  /**
   * Open all tables in list, locks them (all, including derived)
   *
   * @param Pointer to a list of tables for open & locking
   *
   * @retval
   *  false - ok
   * @retval
   *  true  - error
   *
   * @note
   * 
   * The lock will automaticaly be freed by close_thread_tables()
   */
  bool openTablesLock(TableList *tables);

  int open_tables_from_list(TableList **start, uint32_t *counter, uint32_t flags= 0);

  Table *openTableLock(TableList *table_list, thr_lock_type lock_type);
  Table *openTable(TableList *table_list, bool *refresh, uint32_t flags= 0);

  void unlink_open_table(Table *find);
  void drop_open_table(Table *table, const identifier::Table &identifier);
  void close_cached_table(Table *table);

  /* Create a lock in the cache */
  table::Placeholder *table_cache_insert_placeholder(const identifier::Table &identifier);
  bool lock_table_name_if_not_cached(const identifier::Table &identifier, Table **table);

  typedef boost::unordered_map<std::string, message::Table, util::insensitive_hash, util::insensitive_equal_to> TableMessageCache;

  class TableMessages
  {
    TableMessageCache table_message_cache;

  public:
    bool storeTableMessage(const identifier::Table &identifier, message::Table &table_message);
    bool removeTableMessage(const identifier::Table &identifier);
    bool getTableMessage(const identifier::Table &identifier, message::Table &table_message);
    bool doesTableMessageExist(const identifier::Table &identifier);
    bool renameTableMessage(const identifier::Table &from, const identifier::Table &to);
  };

private:
  TableMessages _table_message_cache;

public:
  TableMessages &getMessageCache()
  {
    return _table_message_cache;
  }

  /* Reopen operations */
  bool reopen_tables();
  bool close_cached_tables(TableList *tables, bool wait_for_refresh, bool wait_for_placeholders);

  void wait_for_condition(boost::mutex &mutex, boost::condition_variable_any &cond);
  int setup_conds(TableList *leaves, COND **conds);
  int lock_tables(TableList *tables, uint32_t count, bool *need_reopen);

  drizzled::util::Storable *getProperty(const std::string &arg)
  {
    return life_properties[arg];
  }

  template<class T>
  bool setProperty(const std::string &arg, T *value)
  {
    life_properties[arg]= value;

    return true;
  }

  /**
    Return the default storage engine

    @param getDefaultStorageEngine()

    @return
    pointer to plugin::StorageEngine
  */
  plugin::StorageEngine *getDefaultStorageEngine()
  {
    if (variables.storage_engine)
      return variables.storage_engine;
    return global_system_variables.storage_engine;
  }

  void get_xid(DRIZZLE_XID *xid); // Innodb only

  table::Singular *getInstanceTable();
  table::Singular *getInstanceTable(List<CreateField> &field_list);

private:
  bool resetUsage()
  {
    if (getrusage(RUSAGE_THREAD, &usage))
    {
      return false;
    }

    return true;
  }
public:

  void setUsage(bool arg)
  {
    use_usage= arg;
  }

  const struct rusage &getUsage()
  {
    return usage;
  }

  catalog::Instance::const_reference catalog() const
  {
    return *(_catalog.get());
  }

  catalog::Instance::reference catalog()
  {
    return *(_catalog.get());
  }

private:
  catalog::Instance::shared_ptr _catalog;

  // This lives throughout the life of Session
  bool use_usage;
  PropertyMap life_properties;
  std::vector<table::Singular *> temporary_shares;
  struct rusage usage;
};

class Join;

#define ESCAPE_CHARS "ntrb0ZN" // keep synchronous with READ_INFO::unescape

} /* namespace drizzled */

/** @TODO why is this in the middle of the file */
#include <drizzled/select_to_file.h>
#include <drizzled/select_export.h>
#include <drizzled/select_dump.h>
#include <drizzled/select_insert.h>
#include <drizzled/select_create.h>
#include <drizzled/tmp_table_param.h>
#include <drizzled/select_union.h>
#include <drizzled/select_subselect.h>
#include <drizzled/select_singlerow_subselect.h>
#include <drizzled/select_max_min_finder_subselect.h>
#include <drizzled/select_exists_subselect.h>

namespace drizzled
{

/**
 * A structure used to describe sort information
 * for a field or item used in ORDER BY.
 */
class SortField 
{
public:
  Field *field;	/**< Field to sort */
  Item	*item; /**< Item if not sorting fields */
  size_t length; /**< Length of sort field */
  uint32_t suffix_length; /**< Length suffix (0-4) */
  Item_result result_type; /**< Type of item */
  bool reverse; /**< if descending sort */
  bool need_strxnfrm;	/**< If we have to use strxnfrm() */

  SortField() :
    field(0),
    item(0),
    length(0),
    suffix_length(0),
    result_type(STRING_RESULT),
    reverse(0),
    need_strxnfrm(0)
  { }

};

} /* namespace drizzled */

/** @TODO why is this in the middle of the file */

#include <drizzled/table_ident.h>
#include <drizzled/user_var_entry.h>
#include <drizzled/unique.h>
#include <drizzled/var.h>
#include <drizzled/select_dumpvar.h>

namespace drizzled
{

/* Bits in sql_command_flags */

enum sql_command_flag_bits 
{
  CF_BIT_CHANGES_DATA,
  CF_BIT_HAS_ROW_COUNT,
  CF_BIT_STATUS_COMMAND,
  CF_BIT_SHOW_TABLE_COMMAND,
  CF_BIT_WRITE_LOGS_COMMAND,
  CF_BIT_SIZE
};

static const std::bitset<CF_BIT_SIZE> CF_CHANGES_DATA(1 << CF_BIT_CHANGES_DATA);
static const std::bitset<CF_BIT_SIZE> CF_HAS_ROW_COUNT(1 << CF_BIT_HAS_ROW_COUNT);
static const std::bitset<CF_BIT_SIZE> CF_STATUS_COMMAND(1 << CF_BIT_STATUS_COMMAND);
static const std::bitset<CF_BIT_SIZE> CF_SHOW_TABLE_COMMAND(1 << CF_BIT_SHOW_TABLE_COMMAND);
static const std::bitset<CF_BIT_SIZE> CF_WRITE_LOGS_COMMAND(1 << CF_BIT_WRITE_LOGS_COMMAND);

namespace display  {
const std::string &type(drizzled::Session::global_read_lock_t type);
size_t max_string_length(drizzled::Session::global_read_lock_t type);
} /* namespace display */

} /* namespace drizzled */

#endif /* DRIZZLED_SESSION_H */