123456789_123456789_123456789_123456789_123456789_

Class: ActiveRecord::ConnectionAdapters::AbstractAdapter

Relationships & Source Files
Namespace Children
Classes:
Extension / Inclusion / Inheritance Descendants
Subclasses:
Super Chains via Extension / Inclusion / Inheritance
Class Chain:
Instance Chain:
Inherits: Object
Defined in: activerecord/lib/active_record/connection_adapters/abstract_adapter.rb

Overview

Active Record Abstract Adapter

Active Record supports multiple database systems. AbstractAdapter and related classes form the abstraction layer which makes this possible. An AbstractAdapter represents a connection to a database, and provides an abstract interface for database-specific functionality such as establishing a connection, escaping values, building the right SQL fragments for :offset and :limit options, etc.

All the concrete database adapters follow the interface laid down in this class. ActiveRecord::Base.lease_connection returns an AbstractAdapter object, which you can use.

Most of the methods in the adapter are useful during migrations. Most notably, the instance methods provided by SchemaStatements are very useful.

Constant Summary

::ActiveSupport::Callbacks - Included

CALLBACK_FILTER_TYPES

SchemaStatements - Included

EMPTY_METADATA

DatabaseStatements - Included

DEFAULT_INSERT_VALUE, HIGH_PRECISION_CURRENT_TIMESTAMP, TRANSACTION_ISOLATION_LEVELS

::ActiveSupport::Callbacks - Attributes & Methods

Class Attribute Summary

Class Method Summary

Instance Attribute Summary

DatabaseStatements - Included

SchemaStatements - Included

Instance Method Summary

Savepoints - Included

DatabaseLimits - Included

#index_name_length

Returns the maximum length of an index name.

#table_alias_length

Returns the maximum length of a table alias.

#table_name_length

Returns the maximum length of a table name.

#bind_params_length, #max_identifier_length

Quoting - Included

#quote

Quotes the column value to help prevent SQL injection attacks.

#quote_column_name

Quotes the column name.

#quote_string

Quotes a string, escaping any ' (single quote) and \ (backslash) characters.

#quote_table_name

Quotes the table name.

#quote_table_name_for_assignment

Override to return the quoted table name for assignment.

#quoted_date

Quote date/time values for use in SQL input.

#quoted_false, #quoted_true,
#type_cast

Cast a value to a type that the database understands.

#unquoted_false, #unquoted_true,
#cast_bound_value

Cast a value to be used as a bound parameter of unknown type.

#lookup_cast_type, #quote_default_expression, #quoted_binary, #quoted_time, #sanitize_as_sql_comment, #type_casted_binds

DatabaseStatements - Included

#add_transaction_record

Register a record with the current transaction so that its after_commit and after_rollback callbacks can be called.

#begin_db_transaction

Begins the transaction (and turns off auto-committing).

#begin_isolated_db_transaction

Begins the transaction with the isolation level set.

#commit_db_transaction

Commits the transaction (and turns on auto-committing).

#create
#default_sequence_name,
#delete

Executes the delete statement and returns the number of rows affected.

#disable_lazy_transactions!, #empty_insert_statement_value, #enable_lazy_transactions!,
#exec_delete

Executes delete sql statement in the context of this connection using binds as the bind substitutes.

#exec_insert

Executes insert sql statement in the context of this connection using binds as the bind substitutes.

#exec_query

Executes sql statement in the context of this connection using binds as the bind substitutes.

#exec_update

Executes update sql statement in the context of this connection using binds as the bind substitutes.

#execute

Executes the SQL statement in the context of this connection and returns the raw result from the connection adapter.

#high_precision_current_timestamp

Returns an ::Arel SQL literal for the CURRENT_TIMESTAMP for usage with arbitrary precision date/time columns.

#initialize,
#insert

Executes an INSERT query and returns the new record's ID.

#insert_fixture

Inserts the given fixture into the table.

#insert_fixtures_set,
#reset_isolation_level

Hook point called after an isolated DB transaction is committed or rolled back.

#reset_sequence!

Set the sequence to the max value of the table's column.

#restart_db_transaction,
#rollback_db_transaction

Rolls back the transaction (and turns on auto-committing).

#rollback_to_savepoint,
#select_all

Returns an ::ActiveRecord::Result instance.

#select_one

Returns a record hash with the column names as keys and column values as values.

#select_rows

Returns an array of arrays containing the field values.

#select_value

Returns a single value from a record.

#select_values

Returns an array of the values of the first column in a select:

#to_sql,
#transaction

Runs the given block in a database transaction, and returns the result of the block.

#transaction_isolation_levels,
#truncate

Executes the truncate statement.

#update

Executes the update statement and returns the number of rows affected.

#write_query?

Determines whether the SQL statement is a write query.

#affected_rows, #apply_returning_to!, #arel_from_relation, #build_delete_from_statements, #build_fixture_sql, #build_fixture_statements, #build_truncate_statement, #build_truncate_statements,
#cast_result

Receive a native adapter result object and returns an ::ActiveRecord::Result object.

#collect_warnings, #combine_multi_statements, #extract_table_ref_from_insert_sql, #internal_build_intent, #last_inserted_id, #perform_query, #primary_key_for_insert, #returning_column_values, #single_value_from_rows, #table_ref_for_insert, #_exec_insert, #begin_deferred_transaction,
#cacheable_query

This is used in the ::ActiveRecord::StatementCache object.

#default_insert_value, #empty_all_tables, #exec_insert_all, #exec_restart_db_transaction, #exec_rollback_db_transaction,
#execute_batch

Executes SQL statements in the context of this connection without returning a result.

#execute_intent

Lowest-level abstract execution of a query, called only from the intent itself.

#explain, #finalize_remaining_intents, #finish_intent_log, #handle_warnings, #perform_sync_attempt, #query_all, #query_command, #query_one, #query_rows, #query_value, #query_values, #reset_transaction, #start_intent_log, #to_sql_and_binds, #truncate_tables,
#update_with_result

Executes the update statement and returns an ::ActiveRecord::Result Some adapters support the returning keyword argument.

#with_yaml_fallback

::ActiveRecord::Fixture value is quoted by ::Arel, however scalar values are not quotable.

SchemaStatements - Included

#add_belongs_to
#add_check_constraint

Adds a new check constraint to the table.

#add_column

Add a new type column named column_name to table_name.

#add_foreign_key

Adds a new foreign key.

#add_index

Adds a new index to the table.

#add_reference

Adds a reference.

#add_timestamps

Adds timestamps (created_at and updated_at) columns to table_name.

#assume_migrated_upto_version,
#build_create_table_definition

Returns a TableDefinition object containing information about the table that would be created if the same arguments were passed to #create_table.

#change_column

Changes the column's definition according to the new options.

#change_column_comment

Changes the comment for a column or removes it if nil.

#change_column_default

Sets a new default value for a column:

#change_column_null

Sets or removes a NOT NULL constraint on a column.

#change_foreign_key

Changes an existing foreign key on a table.

#change_table

A block for changing columns in table.

#change_table_comment

Changes the comment for a table or removes it if nil.

#check_constraint_exists?

Checks to see if a check constraint exists on a table for a given check constraint definition.

#check_constraints

Returns an array of check constraints for the given table, or a ::Hash of them keyed by table name when given an ::Array of tables.

#column_exists?

Checks to see if a column exists in a given table.

#columns

Returns an array of Column objects for the given table, or a ::Hash of them keyed by table name when given an ::Array of tables.

#create_join_table

Creates a new join table with the name created using the lexical order of the first two arguments.

#create_table

Creates a new table with the name table_name.

#data_source_exists?

Checks to see if the data source name exists on the database.

#data_sources

Returns the relation names usable to back Active Record models.

#disable_index

Prevents an index from being used by queries.

#drop_join_table

Drops the join table specified by the given arguments.

#drop_table

Drops a table or tables from the database.

#enable_index

Enables an index to be used by queries.

#foreign_key_exists?

Checks to see if a foreign key exists on a table for a given foreign key definition.

#foreign_keys

Returns an array of foreign keys for the given table, or a ::Hash of them keyed by table name when given an ::Array of tables.

#index_exists?

Checks to see if an index exists on a table for a given index definition.

#index_name_exists?

Verifies the existence of an index with a given name.

#indexes

Returns an array of indexes for the given table, or a ::Hash of them keyed by table name when given an ::Array of tables.

#max_index_name_size

Returns the maximum length of an index name in bytes.

#native_database_types

Returns a hash of mappings from the abstract data types to the native database types.

#options_include_default?,
#primary_key

Returns just a table's primary key.

#remove_belongs_to
#remove_check_constraint

Removes the given check constraint from the table.

#remove_column

Removes the column from the table definition.

#remove_columns

Removes the given columns from the table definition.

#remove_foreign_key

Removes the given foreign key from the table.

#remove_index

Removes the given index from the table.

#remove_reference

Removes the reference(s).

#remove_timestamps

Removes the timestamp columns (created_at and updated_at) from the table definition.

#rename_column

Renames a column.

#rename_index

Renames an index.

#rename_table

Renames a table.

#table_alias_for

Truncates a table alias according to the limits of the current adapter.

#table_comment

Returns the table comment that's stored in database metadata.

#table_exists?

Checks to see if the table table_name exists on the database.

#table_options

Returns the options the given table was created with, or a ::Hash of them keyed by table name when given an ::Array of tables.

#tables

Returns an array of table names defined in the database.

#view_exists?

Checks to see if the view view_name exists on the database.

#views

Returns an array of view names defined in the database.

#add_index_sort_order,
#add_options_for_index_columns

Overridden by the MySQL adapter for supporting index lengths and by the PostgreSQL adapter for supporting operator classes.

#can_remove_index_by_name?, #change_column_comment_sql, #change_index_comment_sql, #change_table_comment_sql, #check_constraint_for, #check_constraint_for!, #check_constraint_name, #column_options_keys, #create_alter_table, #create_index_definition, #create_table_definition, #data_source_sql, #drop_table_sql, #execute_alter_table,
#expression_column_name?

Try to identify whether the given column name is an expression.

#extract_foreign_key_action,
#extract_new_comment_value
#extract_new_default_value,
#fetch_by_schema

One read filters by one schema, so tables naming different schemas are read a schema at a time.

#fetch_column_definitions, #fetch_table_options, #fetch_type_metadata, #foreign_key_for, #foreign_key_for!, #foreign_key_name, #generate_index_name, #has_expression_column_name?, #index_column_names, #index_name_for_remove, #index_name_options, #insert_versions_sql, #options_for_index_columns, #quoted_scope, #quoted_table_names, #reference_name_for_table, #rename_column_indexes, #rename_table_indexes, #rows_for, #strip_table_name_prefix_and_suffix, #valid_index_options, #validate_change_column_null_argument!, #validate_create_table_options!, #validate_index_length!, #validate_table_length!, #add_columns, #add_index_options,
#build_alter_table_definition

Returns an AlterTable object containing information about changes that would be made to table_name.

#build_create_index_definition

Builds a CreateIndexDefinition object.

#build_create_join_table_definition

Builds a TableDefinition object for a join table.

#bulk_change_table, #check_constraint_options,
#columns_for_distinct

Given a set of columns and an ORDER BY clause, returns the columns for a SELECT DISTINCT.

#create_schema_dumper, #distinct_relation_for_primary_key, #dump_schema_versions, #foreign_key_column_for, #foreign_key_options, #index_algorithm, #index_name, #internal_string_options_for_primary_key, #quoted_columns_for_index, #remove_constraint,
#schema_creation

Returns an instance of SchemaCreation, which can be used to visit a schema definition object and return DDL.

#type_to_sql, #update_table_definition, #valid_column_definition_options, #valid_primary_key_options, #valid_table_definition_options

::ActiveSupport::Callbacks - Included

#run_callbacks

Runs the callbacks for the given event.

#halted_callback_hook

A hook invoked every time a before callback is halted.

Constructor Details

.new(config_or_deprecated_connection, deprecated_logger = nil, deprecated_connection_options = nil, deprecated_config = nil) ⇒ AbstractAdapter

This method is for internal use only.
[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 145

def initialize(config_or_deprecated_connection, deprecated_logger = nil, deprecated_connection_options = nil, deprecated_config = nil) # :nodoc:
  super()

  @raw_connection = nil
  @unconfigured_connection = nil
  @connected_since = nil

  if config_or_deprecated_connection.is_a?(Hash)
    @config = config_or_deprecated_connection.symbolize_keys
    @logger = ActiveRecord::Base.logger

    if deprecated_logger || deprecated_connection_options || deprecated_config
      raise ArgumentError, "when initializing an Active Record adapter with a config hash, that should be the only argument"
    end
  else
    # Soft-deprecated for now; we'll probably warn in future.

    @unconfigured_connection = config_or_deprecated_connection
    @connected_since = Process.clock_gettime(Process::CLOCK_MONOTONIC)
    @logger = deprecated_logger || ActiveRecord::Base.logger
    if deprecated_config
      @config = (deprecated_config || {}).symbolize_keys
      @connection_parameters = deprecated_connection_options
    else
      @config = (deprecated_connection_options || {}).symbolize_keys
      @connection_parameters = nil
    end
  end

  @owner = nil
  @pinned = false
  @pool = ActiveRecord::ConnectionAdapters::NullPool.new
  @idle_since = Process.clock_gettime(Process::CLOCK_MONOTONIC)
  @allow_preconnect = false
  @visitor = arel_visitor
  @statements = build_statement_pool
  self.lock_thread = nil

  @prepared_statements = !ActiveRecord.disable_prepared_statements && self.class.type_cast_config_to_boolean(
    @config.fetch(:prepared_statements) { default_prepared_statements }
  )

  @advisory_locks_enabled = self.class.type_cast_config_to_boolean(
    @config.fetch(:advisory_locks, true)
  )

  @sql_notifications = self.class.type_cast_config_to_boolean(
    @config.fetch(:sql_notifications, true)
  )

  @default_timezone = self.class.validate_default_timezone(@config[:default_timezone])

  @raw_connection_dirty = false
  @last_activity = nil
  @verified = false
  @needs_reconnect = false
  @unfinalized_intents = []

  @pool_jitter = rand * max_jitter
end

Class Attribute Details

.__callbacks (rw)

[ GitHub ]

  
# File 'activesupport/lib/active_support/callbacks.rb', line 70

class_attribute :__callbacks, instance_writer: false, instance_predicate: false, default: {}.freeze

.migration_strategy (rw)

Allows configuration of migration strategy per adapter type. When set on a specific adapter class (e.g., PostgreSQLAdapter), all migrations using that adapter will use the specified strategy instead of the global ActiveRecord.migration_strategy.

ActiveRecord::ConnectionAdapters::PostgreSQLAdapter.migration_strategy = CustomPostgresStrategy
ActiveRecord::ConnectionAdapters::Mysql2Adapter.migration_strategy = CustomMySQLStrategy
[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 44

class_attribute :migration_strategy, instance_writer: false

.migration_strategy?Boolean (rw)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 44

class_attribute :migration_strategy, instance_writer: false

Class Method Details

.build_read_query_regexp(*parts)

This method is for internal use only.
[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 106

def self.build_read_query_regexp(*parts) # :nodoc:
  parts += DEFAULT_READ_QUERY
  parts = parts.map { |part| /#{part}/i }
  /\A(?:[(\s]|#{COMMENT_REGEX})*#{Regexp.union(*parts)}/
end

.database_exists?(config) ⇒ Boolean

Does the database for this adapter exist?

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 422

def self.database_exists?(config)
  new(config).database_exists?
end

.dbconsole(config, options = {})

Opens a database console session.

Raises:

  • (NotImplementedError)
[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 141

def self.dbconsole(config, options = {})
  raise NotImplementedError.new("#{self} should define `dbconsole` that accepts a db config and options to implement connecting to the db console")
end

.extended_type_map(default_timezone:)

This method is for internal use only.
[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 979

def extended_type_map(default_timezone:) # :nodoc:
  Type::TypeMap.new(self::TYPE_MAP).tap do |m|
    register_class_with_precision m, %r(\A[^\(]*time)i, Type::Time, timezone: default_timezone
    register_class_with_precision m, %r(\A[^\(]*datetime)i, Type::DateTime, timezone: default_timezone
    m.alias_type %r(\A[^\(]*timestamp)i, "datetime"
  end
end

.extract_limit(sql_type) (private)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 1043

def extract_limit(sql_type)
  $1.to_i if sql_type =~ /\((.*)\)/
end

.extract_precision(sql_type) (private)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 1039

def extract_precision(sql_type)
  $1.to_i if sql_type =~ /\((\d)(,\d)?\)/
end

.extract_scale(sql_type) (private)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 1032

def extract_scale(sql_type)
  case sql_type
  when /\((\d+)\)/ then 0
  when /\((\d)(,(\d))\)/ then $3.to_i
  end
end

.find_cmd_and_exec(commands, *args)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 112

def self.find_cmd_and_exec(commands, *args) # :doc:
  commands = Array(commands)

  dirs_on_path = ENV["PATH"].to_s.split(File::PATH_SEPARATOR)
  unless (ext = RbConfig::CONFIG["EXEEXT"]).empty?
    commands = commands.map { |cmd| "#{cmd}#{ext}" }
  end

  full_path_command = nil
  found = commands.detect do |cmd|
    dirs_on_path.detect do |path|
      full_path_command = File.join(path, cmd)
      begin
        stat = File.stat(full_path_command)
      rescue SystemCallError
      else
        stat.file? && stat.executable?
      end
    end
  end

  if found
    exec full_path_command, *args
  else
    abort("Couldn't find database client: #{commands.join(', ')}. Check your $PATH and try again.")
  end
end

.initialize_type_map(m) (private)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 992

def initialize_type_map(m)
  register_class_with_limit m, %r(boolean)i,       Type::Boolean
  register_class_with_limit m, %r(char)i,          Type::String
  register_class_with_limit m, %r(binary)i,        Type::Binary
  register_class_with_limit m, %r(text)i,          Type::Text
  register_class_with_precision m, %r(date)i,      Type::Date
  register_class_with_precision m, %r(time)i,      Type::Time
  register_class_with_precision m, %r(datetime)i,  Type::DateTime
  register_class_with_limit m, %r(float)i,         Type::Float
  register_class_with_limit m, %r(int)i,           Type::Integer

  m.alias_type %r(blob)i,      "binary"
  m.alias_type %r(clob)i,      "text"
  m.alias_type %r(timestamp)i, "datetime"
  m.alias_type %r(numeric)i,   "decimal"
  m.alias_type %r(number)i,    "decimal"
  m.alias_type %r(double)i,    "float"

  m.register_type %r(^json)i, Type::Json.new.freeze

  m.register_type(%r(decimal)i) do |sql_type|
    scale = extract_scale(sql_type)
    precision = extract_precision(sql_type)

    if scale == 0
      # FIXME: Remove this class as well
      Type::DecimalWithoutScale.new(precision: precision).freeze
    else
      Type::Decimal.new(precision: precision, scale: scale).freeze
    end
  end
end

.register_class_with_limit(mapping, key, klass) (private)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 1025

def register_class_with_limit(mapping, key, klass)
  mapping.register_type(key) do |*args|
    limit = extract_limit(args.last)
    klass.new(limit: limit).freeze
  end
end

.register_class_with_precision(mapping, key, klass, **kwargs)

This method is for internal use only.
[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 972

def register_class_with_precision(mapping, key, klass, **kwargs) # :nodoc:
  mapping.register_type(key) do |*args|
    precision = extract_precision(args.last)
    klass.new(precision: precision, **kwargs).freeze
  end
end

.type_cast_config_to_boolean(config)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 85

def self.type_cast_config_to_boolean(config)
  if config == "false"
    false
  else
    config
  end
end

.type_cast_config_to_integer(config)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 75

def self.type_cast_config_to_integer(config)
  if config.is_a?(Integer)
    config
  elsif SIMPLE_INT.match?(config)
    config.to_i
  else
    config
  end
end

.valid_type?(type) ⇒ Boolean

This method is for internal use only.
[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 987

def valid_type?(type) # :nodoc:
  !native_database_types[type].nil?
end

.validate_default_timezone(config)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 93

def self.validate_default_timezone(config)
  case config
  when nil
  when "utc", "local"
    config.to_sym
  else
    raise ArgumentError, "default_timezone must be either 'utc' or 'local'"
  end
end

Instance Attribute Details

#__callbacks (readonly)

[ GitHub ]

  
# File 'activesupport/lib/active_support/callbacks.rb', line 70

class_attribute :__callbacks, instance_writer: false, instance_predicate: false, default: {}.freeze

#active?Boolean (readonly)

Checks whether the connection to the database is still active. This includes checking whether the database is actually capable of responding, i.e. whether the connection isn't stale.

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 737

def active?
end

#advisory_locks_enabled?Boolean (readonly)

This method is for internal use only.
[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 684

def advisory_locks_enabled? # :nodoc:
  supports_advisory_locks? && @advisory_locks_enabled
end

#allow_preconnect (rw)

This method is for internal use only.
[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 59

attr_reader :allow_preconnect # :nodoc:

#allow_preconnect=(value) (rw)

This method is for internal use only.
[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 69

def allow_preconnect=(value) # :nodoc:
  @lock.synchronize do
    @allow_preconnect = value
  end
end

#async_enabled?Boolean (readonly)

This method is for internal use only.
[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 643

def async_enabled? # :nodoc:
  supports_concurrent_connections? &&
    !ActiveRecord.async_query_executor.nil? && !pool.async_executor.nil?
end

#connected?Boolean (readonly)

Checks whether the connection to the database was established. This doesn't include checking whether the database is actually capable of responding, i.e. whether the connection is stale.

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 730

def connected?
  !@raw_connection.nil?
end

#database_exists?Boolean (readonly)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 426

def database_exists?
  connect!
  true
rescue ActiveRecord::NoDatabaseError
  false
end

#in_use? (readonly)

Alias for #owner.

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 61

alias :in_use? :owner

#lock (readonly)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 58

attr_reader :visitor, :owner, :logger, :lock

#lock_thread=(lock_thread) (writeonly)

This method is for internal use only.
[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 213

def lock_thread=(lock_thread) # :nodoc:
  @lock =
  case lock_thread
  when Thread
    ActiveSupport::Concurrency::ThreadMonitor.new
  when Fiber
    ::Monitor.new
  else
    ActiveSupport::Concurrency::NullLock
  end
end

#logger (readonly)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 58

attr_reader :visitor, :owner, :logger, :lock

#migration_strategy (readonly)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 44

class_attribute :migration_strategy, instance_writer: false

#migration_strategy?Boolean (readonly)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 44

class_attribute :migration_strategy, instance_writer: false

#needs_reconnect?Boolean (readonly)

This method is for internal use only.
[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 890

def needs_reconnect? # :nodoc:
  @needs_reconnect
end

#owner (readonly) Also known as: #in_use?

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 58

attr_reader :visitor, :owner, :logger, :lock

#pinned (rw)

This method is for internal use only.
[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 60

attr_accessor :pinned # :nodoc:

#pool (rw)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 57

attr_reader :pool

#pool=(value) (rw)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 63

def pool=(value)
  return if value.eql?(@pool)
  @schema_cache = nil
  @pool = value
end

#prepared_statements (readonly)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 268

alias :prepared_statements :prepared_statements?

#preventing_writes?Boolean (readonly)

Determines whether writes are currently being prevented.

Returns true if the connection is a replica or returns the value of current_preventing_writes.

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 258

def preventing_writes?
  return true if replica?
  return false if connection_descriptor.nil?

  connection_descriptor.current_preventing_writes
end

#reconnect_can_restore_state?Boolean (readonly, private)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 1087

def reconnect_can_restore_state?
  transaction_manager.restorable? && !@raw_connection_dirty
end

#replica?Boolean (readonly)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 230

def replica?
  @config[:replica] || false
end

#requires_reloading?Boolean (readonly)

Returns true if its required to reload the connection between requests for development mode.

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 836

def requires_reloading?
  false
end

#savepoint_errors_invalidate_transactions?Boolean (readonly)

Do TransactionRollbackErrors on savepoints affect the parent transaction?

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 450

def savepoint_errors_invalidate_transactions?
  false
end

#sql_notifications?Boolean (readonly)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 274

def sql_notifications?
  @sql_notifications
end

#supports_advisory_locks?Boolean (readonly)

Does this adapter support application-enforced advisory locking?

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 459

def supports_advisory_locks?
  false
end

#supports_bulk_alter?Boolean (readonly)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 439

def supports_bulk_alter?
  false
end

#supports_check_constraints?Boolean (readonly)

Does this adapter support creating check constraints?

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 536

def supports_check_constraints?
  false
end

#supports_comments?Boolean (readonly)

Does this adapter support metadata comments on database objects (tables, columns, indexes)?

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 571

def supports_comments?
  false
end

#supports_comments_in_create?Boolean (readonly)

Can comments for tables, columns, and indexes be specified in create/alter table statements?

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 576

def supports_comments_in_create?
  false
end

#supports_common_table_expressions?Boolean (readonly)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 595

def supports_common_table_expressions?
  false
end

#supports_concurrent_connections?Boolean (readonly)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 623

def supports_concurrent_connections?
  true
end

#supports_datetime_with_precision?Boolean (readonly)

Does this adapter support datetime with precision?

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 561

def supports_datetime_with_precision?
  false
end

#supports_ddl_transactions?Boolean (readonly)

Does this adapter support DDL rollbacks in transactions? That is, would CREATE TABLE or ALTER TABLE get rolled back by a transaction?

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 435

def supports_ddl_transactions?
  false
end

#supports_deferrable_constraints?Boolean (readonly)

Does this adapter support creating deferrable constraints?

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 526

def supports_deferrable_constraints?
  false
end

#supports_disabling_indexes?Boolean (readonly)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 631

def supports_disabling_indexes?
  false
end

#supports_enforced_foreign_keys?Boolean (readonly)

Does this adapter support NOT ENFORCED foreign key constraints?

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 531

def supports_enforced_foreign_keys?
  false
end

#supports_exclusion_constraints?Boolean (readonly)

Does this adapter support creating exclusion constraints?

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 541

def supports_exclusion_constraints?
  false
end

#supports_explain?Boolean (readonly)

Does this adapter support explain?

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 495

def supports_explain?
  false
end

#supports_expression_index?Boolean (readonly)

Does this adapter support expression indices?

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 490

def supports_expression_index?
  false
end

#supports_extensions?Boolean (readonly)

Does this adapter support database extensions?

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 505

def supports_extensions?
  false
end

#supports_foreign_keys?Boolean (readonly)

Does this adapter support creating foreign key constraints?

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 516

def supports_foreign_keys?
  false
end

#supports_foreign_tables?Boolean (readonly)

Does this adapter support foreign/external tables?

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 586

def supports_foreign_tables?
  false
end

#supports_index_include?Boolean (readonly)

Does this adapter support including non-key columns?

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 485

def supports_index_include?
  false
end

#supports_index_sort_order?Boolean (readonly)

Does this adapter support index sort order?

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 475

def supports_index_sort_order?
  false
end

#supports_indexes_in_create?Boolean (readonly)

Does this adapter support creating indexes in the same statement as creating the table?

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 511

def supports_indexes_in_create?
  false
end

#supports_insert_conflict_target?Boolean (readonly)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 619

def supports_insert_conflict_target?
  false
end

#supports_insert_on_duplicate_skip?Boolean (readonly)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 611

def supports_insert_on_duplicate_skip?
  false
end

#supports_insert_on_duplicate_update?Boolean (readonly)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 615

def supports_insert_on_duplicate_update?
  false
end

#supports_insert_returning?Boolean (readonly)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 603

def supports_insert_returning?
  false
end

#supports_json?Boolean (readonly)

Does this adapter support JSON data type?

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 566

def supports_json?
  false
end

#supports_lazy_transactions?Boolean (readonly)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 599

def supports_lazy_transactions?
  false
end

#supports_materialized_views?Boolean (readonly)

Does this adapter support materialized views?

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 556

def supports_materialized_views?
  false
end

#supports_nulls_not_distinct?Boolean (readonly)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 627

def supports_nulls_not_distinct?
  false
end

#supports_optimizer_hints?Boolean (readonly)

Does this adapter support optimizer hints?

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 591

def supports_optimizer_hints?
  false
end

#supports_partial_index?Boolean (readonly)

Does this adapter support partial indices?

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 480

def supports_partial_index?
  false
end

#supports_partitioned_indexes?Boolean (readonly)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 470

def supports_partitioned_indexes?
  false
end

#supports_restart_db_transaction?Boolean (readonly)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 454

def supports_restart_db_transaction?
  false
end

#supports_savepoints?Boolean (readonly)

Does this adapter support savepoints?

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 444

def supports_savepoints?
  false
end

#supports_transaction_isolation?Boolean (readonly)

Does this adapter support setting the isolation level for a transaction?

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 500

def supports_transaction_isolation?
  false
end

#supports_unique_constraints?Boolean (readonly)

Does this adapter support creating unique constraints?

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 546

def supports_unique_constraints?
  false
end

#supports_update_returning?Boolean (readonly)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 607

def supports_update_returning?
  false
end

#supports_validate_constraints?Boolean (readonly)

Does this adapter support creating invalid constraints?

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 521

def supports_validate_constraints?
  false
end

#supports_views?Boolean (readonly)

Does this adapter support views?

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 551

def supports_views?
  false
end

#supports_virtual_columns?Boolean (readonly)

Does this adapter support virtual columns?

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 581

def supports_virtual_columns?
  false
end

#verified?Boolean (readonly)

This method is for internal use only.
[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 886

def verified? # :nodoc:
  @verified
end

#visitor (readonly)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 58

attr_reader :visitor, :owner, :logger, :lock

Instance Method Details

#adapter_name

Returns the human-readable name of the adapter. Use mixed case - one can always use downcase if needed.

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 417

def adapter_name
  self.class::ADAPTER_NAME
end

#add_enum_value

This method is for internal use only.

This is meant to be implemented by the adapters that support custom enum types

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 669

def add_enum_value(...) # :nodoc:
end

#any_raw_connection (private)

Returns a raw connection for internal use with methods that are known to both be thread-safe and not rely upon actual server communication. This is useful for e.g. string escaping methods.

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 1234

def any_raw_connection
  @raw_connection || valid_raw_connection
end

#arel_visitor (private)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 1353

def arel_visitor
  Arel::Visitors::ToSql.new(self)
end

#attempt_configure_connection (private)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 1371

def attempt_configure_connection
  yield
rescue Exception # Need to handle things such as Timeout::ExitException
  disconnect!
  raise
end

#attempt_retry(exception, budget)

This method is for internal use only.

Consume from budget if exception warrants another attempt. Query retries back off here; connection retries consume the reconnect allowance and rely on the caller to ensure connection readiness before executing again.

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 1060

def attempt_retry(exception, budget) # :nodoc:
  return false unless retryable_failure?(exception, budget) && budget.consume

  if retryable_query_error?(exception)
    backoff(budget.attempts_used)
  else
    budget.reconnect_consumed!
  end

  true
end

#backoff(counter) (private)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 1223

def backoff(counter)
  sleep 0.1 * counter
end

#build_insert_sql(insert)

This method is for internal use only.

Called by ::ActiveRecord::InsertAll, Passed an instance of ::ActiveRecord::InsertAll::Builder, This method implements standard bulk inserts for all databases, but should be overridden by adapters to implement common features with non-standard syntax like handling duplicates or returning values.

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 945

def build_insert_sql(insert) # :nodoc:
  if insert.skip_duplicates? || insert.update_duplicates?
    raise NotImplementedError, "#{self.class} should define `build_insert_sql` to implement adapter-specific logic for handling duplicates during INSERT"
  end

  "INSERT #{insert.into} #{insert.values_list}"
end

#build_retry_budget(allow_retry:, reconnectable:) (private)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 1185

def build_retry_budget(allow_retry:, reconnectable:)
  RetryBudget.new(
    retries: allow_retry ? connection_retries : 0,
    deadline: retry_deadline && Process.clock_gettime(Process::CLOCK_MONOTONIC) + retry_deadline,
    reconnectable: reconnectable
  )
end

#build_statement_pool (private)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 1357

def build_statement_pool
end

#can_perform_case_insensitive_comparison_for?(column) ⇒ Boolean (private)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 926

def can_perform_case_insensitive_comparison_for?(column)
  true
end

#case_insensitive_comparison(attribute, value)

This method is for internal use only.
[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 916

def case_insensitive_comparison(attribute, value) # :nodoc:
  column = column_for_attribute(attribute)

  if can_perform_case_insensitive_comparison_for?(column)
    attribute.lower.eq(attribute.relation.lower(value))
  else
    attribute.eq(value)
  end
end

#case_sensitive_comparison(attribute, value)

This method is for internal use only.
[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 912

def case_sensitive_comparison(attribute, value) # :nodoc:
  attribute.eq(value)
end

#check_all_foreign_keys_valid!

Override to check all foreign key constraints in a database. The adapter should raise a ::ActiveRecord::StatementInvalid if foreign key constraints are not met.

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 722

def check_all_foreign_keys_valid!
end

#check_version

This method is for internal use only.
[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 960

def check_version # :nodoc:
end

#clean!

This method is for internal use only.
[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 878

def clean! # :nodoc:
  _run_checkout_callbacks do
    @raw_connection_dirty = false
    @verified = nil
  end
  self
end

#clear_cache!(new_connection: false)

Clear any caching the database adapter may be doing.

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 823

def clear_cache!(new_connection: false)
  if @statements
    @lock.synchronize do
      if new_connection
        @statements.reset
      else
        @statements.clear
      end
    end
  end
end

#close

Check the connection back in to the connection pool

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 932

def close
  pool.checkin self
end

#collector (private)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 1339

def collector
  if prepared_statements
    Arel::Collectors::Composite.new(
      Arel::Collectors::SQLString.new,
      Arel::Collectors::Bind.new,
    )
  else
    Arel::Collectors::SubstituteBinds.new(
      self,
      Arel::Collectors::SQLString.new,
    )
  end
end

#column_for(table_name, column_name) (private)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 1328

def column_for(table_name, column_name)
  column_name = column_name.to_s
  columns(table_name).detect { |c| c.name == column_name } ||
    raise(ActiveRecordError, "No such column: #{table_name}.#{column_name}")
end

#column_for_attribute(attribute) (private)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 1334

def column_for_attribute(attribute)
  table_name = attribute.relation.name
  schema_cache.columns_hash(table_name)[attribute.name.to_s]
end

#configure_connection (private)

Perform any necessary initialization upon the newly-established connection settings, run queries to configure any application-global "session" variables, etc.

Implementations may assume this method will only be called while holding @lock (or from #initialize).

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 1367

def configure_connection
  check_version
end

#connect!

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 873

def connect!
  verify!
  self
end

#connection_age

This method is for internal use only.

Seconds since this connection was established. nil if not connected; infinity if the connection has been explicitly retired.

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 396

def connection_age # :nodoc:
  if @raw_connection && @connected_since
    Process.clock_gettime(Process::CLOCK_MONOTONIC) - @connected_since
  end
end

#connection_descriptor

This method is for internal use only.
[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 321

def connection_descriptor # :nodoc:
  @pool.connection_descriptor
end

#connection_retries

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 234

def connection_retries
  (@config[:connection_retries] || 1).to_i
end

#create_enum

This method is for internal use only.

This is meant to be implemented by the adapters that support custom enum types

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 657

def create_enum(...) # :nodoc:
end

#create_virtual_table

This method is for internal use only.

This is meant to be implemented by the adapters that support virtual tables

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 677

def create_virtual_table(*) # :nodoc:
end

#database_version Also known as: #postgresql_version

This method is for internal use only.
[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 956

def database_version # :nodoc:
  pool.server_version(self)
end

#default_index_type?(index) ⇒ Boolean

This method is for internal use only.
[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 936

def default_index_type?(index) # :nodoc:
  index.using.nil?
end

#default_prepared_statements (private)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 1378

def default_prepared_statements
  true
end

#default_timezone

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 250

def default_timezone
  @default_timezone || ActiveRecord.default_timezone
end

#disable_extension(name)

This is meant to be implemented by the adapters that support extensions

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 649

def disable_extension(name, **)
end

#disable_referential_integrity

Override to turn off referential integrity while executing &block.

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 715

def disable_referential_integrity
  yield
end

#discard!

Immediately forget this connection ever existed. Unlike disconnect!, this will not communicate with the server.

After calling this method, the behavior of all other methods becomes undefined. This is called internally just before a forked process gets rid of a connection that belonged to its parent.

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 796

def discard!
  # This should be overridden by concrete adapters.
end

#disconnect!

Disconnects from the database if already connected. Otherwise, this method does nothing.

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 778

def disconnect!
  @lock.synchronize do
    clear_cache!(new_connection: true)
    reset_transaction
    @raw_connection_dirty = false
    @connected_since = nil
    @last_activity = nil
    @verified = false
    @needs_reconnect = false
  end
end

#downgrade_connection_after_error(exception)

This method is for internal use only.
[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 1072

def downgrade_connection_after_error(exception) # :nodoc:
  unless retryable_query_error?(exception)
    # Barring a known-retryable error inside the query (regardless of
    # whether we were in a _position_ to retry it), we should infer that
    # there's likely a real problem with the connection.
    @last_activity = nil
    @verified = false

    if retryable_connection_error?(exception)
      @needs_reconnect = true
    end
  end
end

#drop_enum

This method is for internal use only.

This is meant to be implemented by the adapters that support custom enum types

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 661

def drop_enum(...) # :nodoc:
end

#drop_virtual_table

This method is for internal use only.

This is meant to be implemented by the adapters that support virtual tables

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 681

def drop_virtual_table(*) # :nodoc:
end

#enable_extension(name)

This is meant to be implemented by the adapters that support extensions

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 653

def enable_extension(name, **)
end

#ensure_connection_ready(allow_retry:, materialize_transactions:) (private)

Ensure the connection is ready to execute a query. Returns whether reconnect-and-restore is available for retry decisions.

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 1195

def ensure_connection_ready(allow_retry:, materialize_transactions:)
  connect! if !connected? && reconnect_can_restore_state?
  self.materialize_transactions if materialize_transactions

  reconnectable = reconnect_can_restore_state?
  verify! unless skip_verification?(allow_retry:, reconnectable:)
  reconnectable
end

#expire(update_idle = true)

This method is for internal use only.

this method must only be called while holding connection pool's mutex

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 346

def expire(update_idle = true) # :nodoc:
  if in_use?
    if @owner != ActiveSupport::IsolatedExecutionState.context
      raise ActiveRecordError, "Cannot expire connection, " \
        "it is owned by a different thread: #{@owner}. " \
        "Current thread: #{ActiveSupport::IsolatedExecutionState.context}."
    end

    finalize_remaining_intents

    _run_checkin_callbacks do
      @idle_since = Process.clock_gettime(Process::CLOCK_MONOTONIC) if update_idle
      @owner = nil
      enable_lazy_transactions!
      unset_query_cache!
    end
  else
    raise ActiveRecordError, "Cannot expire connection, it is not currently leased."
  end
end

#extended_type_map_key (private)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 1251

def extended_type_map_key
  if @default_timezone
    { default_timezone: @default_timezone }
  end
end

#extensions

A list of extensions, to be filled in by adapters that support them.

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 703

def extensions
  []
end

#force_retirement

This method is for internal use only.

Mark the connection as needing to be retired, as if the age has exceeded the maximum allowed.

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 404

def force_retirement # :nodoc:
  @connected_since &&= -Float::INFINITY
end

#get_advisory_lock(lock_id)

This method is for internal use only.

This is meant to be implemented by the adapters that support advisory locks

Return true if we got the lock, otherwise false

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 692

def get_advisory_lock(lock_id) # :nodoc:
end

#get_database_version

This method is for internal use only.
[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 953

def get_database_version # :nodoc:
end

#index_algorithms

A list of index algorithms, to be filled by adapters that support them.

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 708

def index_algorithms
  {}
end

#inspect

This method is for internal use only.
[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 206

def inspect # :nodoc:
  name_field = " name=#{pool.db_config.name.inspect}" unless pool.db_config.name == "primary"
  shard_field = " shard=#{shard.inspect}" unless shard == :default

  "#<#{self.class.name}:#{'%#016x' % (object_id << 1)} env_name=#{pool.db_config.env_name.inspect}#{name_field} role=#{role.inspect}#{shard_field}>"
end

#instrumenter (private)

This method is for internal use only.
[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 1310

def instrumenter # :nodoc:
  if sql_notifications?
    ActiveSupport::IsolatedExecutionState[:active_record_instrumenter] ||= ActiveSupport::Notifications.instrumenter
  else
    ActiveSupport::Notifications.null_instrumenter
  end
end

#invalidate_transaction(exception) (private)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 1171

def invalidate_transaction(exception)
  return unless exception.is_a?(TransactionRollbackError)
  return unless savepoint_errors_invalidate_transactions?

  current_transaction.invalidate!
end

#lease

this method must only be called while holding connection pool's mutex

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 306

def lease
  if in_use?
    msg = +"Cannot lease connection, "
    if @owner == ActiveSupport::IsolatedExecutionState.context
      msg << "it is already leased by the current thread."
    else
      msg << "it is already in use by a different thread: #{@owner}. " \
             "Current thread: #{ActiveSupport::IsolatedExecutionState.context}."
    end
    raise ActiveRecordError, msg
  end

  @owner = ActiveSupport::IsolatedExecutionState.context
end

#log(sql, name = "SQL", binds = [], type_casted_binds = [], async: false, allow_retry: false, &block) (private)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 1286

def log(sql, name = "SQL", binds = [], type_casted_binds = [], async: false, allow_retry: false, &block)
  ActiveRecord.deprecator.warn(<<-MSG.squish)
    `log` is deprecated and will be removed in Rails 8.3.
    Queries executed through a `QueryIntent` are instrumented automatically.
  MSG

  instrumenter.instrument(
    "sql.active_record",
    sql:               sql,
    name:              name,
    binds:             binds,
    type_casted_binds: type_casted_binds,
    async:             async,
    allow_retry:       allow_retry,
    connection:        self,
    transaction:       current_transaction.user_transaction.presence,
    affected_rows:     0,
    row_count:         0,
    &block
  )
rescue ActiveRecord::StatementInvalid => ex
  raise ex.set_query(sql, binds)
end

#max_jitter

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 226

def max_jitter
  (@config[:pool_jitter] || 0.2).to_f.clamp(MAX_JITTER)
end

#native_database_types

This method is for internal use only.
[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 301

def native_database_types # :nodoc:
  self.class.native_database_types
end

#pool_jitter(duration)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 341

def pool_jitter(duration)
  duration * (1.0 - @pool_jitter)
end

#prefetch_primary_key?(table_name = nil) ⇒ Boolean

Should primary key values be selected from their corresponding sequence before the insert statement? If true, next_sequence_value is called before each insert to set the record's primary key.

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 466

def prefetch_primary_key?(table_name = nil)
  false
end

#prepared_statements?Boolean (readonly) Also known as: #prepared_statements

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 265

def prepared_statements?
  @prepared_statements && !prepared_statements_disabled_cache.include?(object_id)
end

#prepared_statements_disabled_cache

This method is for internal use only.
[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 270

def prepared_statements_disabled_cache # :nodoc:
  ActiveSupport::IsolatedExecutionState[:active_record_prepared_statements_disabled_cache] ||= Set.new
end

#raw_connection

Provides access to the underlying database driver for this adapter. For example, this method returns a Mysql2::Client object in case of Mysql2Adapter, and a PG::Connection object in case of PostgreSQLAdapter.

This is useful for when you need to call a proprietary method such as PostgreSQL's lo_* methods.

Active Record cannot track if the database is getting modified using this client. If that is the case, generally you'll want to invalidate the query cache using ActiveRecord::Base.clear_query_cache.

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 904

def raw_connection
  with_raw_connection do |conn|
    disable_lazy_transactions!
    @raw_connection_dirty = true
    conn
  end
end

#reconnect (private)

Raises:

  • (NotImplementedError)
[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 1227

def reconnect
  raise NotImplementedError.new("#{self.class} should define `reconnect` to implement adapter-specific logic for reconnecting to the database")
end

#reconnect!(restore_transactions: false)

Disconnects from the database if already connected, and establishes a new connection with the database. Implementors should define private #reconnect instead.

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 743

def reconnect!(restore_transactions: false)
  budget = build_retry_budget(allow_retry: true, reconnectable: false)

  @lock.synchronize do
    attempt_configure_connection do
      @allow_preconnect = false

      reconnect
      @needs_reconnect = false

      enable_lazy_transactions!
      @raw_connection_dirty = false
      @last_activity = @connected_since = Process.clock_gettime(Process::CLOCK_MONOTONIC)
      @verified = true
      @allow_preconnect = true

      reset_transaction(restore: restore_transactions) do
        clear_cache!(new_connection: true)
        configure_connection
      end
    rescue => original_exception
      translated_exception = translate_exception_class(original_exception, nil, nil)

      if retryable_connection_error?(translated_exception) && budget.consume
        backoff(budget.attempts_used)
        retry
      end

      raise translated_exception
    end
  end
end

#release_advisory_lock(lock_id)

This method is for internal use only.

This is meant to be implemented by the adapters that support advisory locks.

Return true if we released the lock, otherwise false

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 699

def release_advisory_lock(lock_id) # :nodoc:
end

#rename_enum

This method is for internal use only.

This is meant to be implemented by the adapters that support custom enum types

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 665

def rename_enum(...) # :nodoc:
end

#rename_enum_value

This method is for internal use only.

This is meant to be implemented by the adapters that support custom enum types

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 673

def rename_enum_value(...) # :nodoc:
end

#reset!

Reset the state of this connection, directing the DBMS to clear transactions and other connection-related server-side state. Usually a database-dependent operation.

If a database driver or protocol does not support such a feature, implementors may alias this to #reconnect!. Otherwise, implementors should call super immediately after resetting the connection (and while still holding @lock).

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 808

def reset!
  attempt_configure_connection do
    clear_cache!(new_connection: true)
    reset_transaction
    configure_connection
  end
end

#retry_deadline

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 242

def retry_deadline
  if @config[:retry_deadline]
    @config[:retry_deadline].to_f
  else
    nil
  end
end

#retryable_connection_error?(exception) ⇒ Boolean (private)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 1166

def retryable_connection_error?(exception)
  (exception.is_a?(ConnectionNotEstablished) && !exception.is_a?(ConnectionNotDefined)) ||
    exception.is_a?(ConnectionFailed)
end

#retryable_failure?(exception, budget) ⇒ Boolean

This method is for internal use only.
[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 1051

def retryable_failure?(exception, budget) # :nodoc:
  budget&.available? &&
    (retryable_query_error?(exception) ||
      (budget.reconnectable? && retryable_connection_error?(exception)))
end

#retryable_query_error?(exception) ⇒ Boolean (private)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 1178

def retryable_query_error?(exception)
  # We definitely can't retry if we were inside an invalidated transaction.
  return false if current_transaction.invalidated?

  exception.is_a?(Deadlocked) || exception.is_a?(LockWaitTimeout)
end

#return_value_after_insert?(column) ⇒ Boolean

This method is for internal use only.
[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 635

def return_value_after_insert?(column) # :nodoc:
  column.auto_populated_on_insert?
end

#return_value_after_update?(column) ⇒ Boolean

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 639

def return_value_after_update?(column)
  column.auto_populated_on_update?
end

#role

The role (e.g. :writing) for the current connection. In a non-multi role application, :writing is returned.

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 327

def role
  @pool.role
end

#schema_cache

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 337

def schema_cache
  @pool.schema_cache || (@schema_cache ||= BoundSchemaReflection.for_lone_connection(@pool.schema_reflection, self))
end

#schema_version

Returns the version identifier of the schema currently available in the database. This is generally equal to the number of the highest- numbered migration that has been executed, or 0 if no schema information is present / the database is empty.

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 967

def schema_version
  pool.migration_context.current_version
end

#seconds_idle

This method is for internal use only.

Seconds since this connection was returned to the pool

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 381

def seconds_idle # :nodoc:
  return 0 if in_use?
  Process.clock_gettime(Process::CLOCK_MONOTONIC) - @idle_since
end

#seconds_since_last_activity

This method is for internal use only.

Seconds since this connection last communicated with the server

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 387

def seconds_since_last_activity # :nodoc:
  if @raw_connection && @last_activity
    Process.clock_gettime(Process::CLOCK_MONOTONIC) - @last_activity
  end
end

#shard

The shard (e.g. :default) for the current connection. In a non-sharded application, :default is returned.

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 333

def shard
  @pool.shard
end

#skip_verification?(allow_retry:, reconnectable:) ⇒ Boolean (private)

Decide whether the connection can be used without verification.

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 1205

def skip_verification?(allow_retry:, reconnectable:)
  if @verified && connected? && !@needs_reconnect
    true
  elsif !@needs_reconnect && (last_activity = seconds_since_last_activity) && last_activity < verify_timeout
    true
  elsif reconnectable
    if @needs_reconnect
      false
    elsif allow_retry
      true
    else
      false
    end
  else
    true
  end
end

#steal!

This method is for internal use only.

this method must only be called while holding connection pool's mutex (and a desire for segfaults)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 368

def steal! # :nodoc:
  if in_use?
    if @owner != ActiveSupport::IsolatedExecutionState.context
      pool.send :remove_connection_from_thread_cache, self, @owner

      @owner = ActiveSupport::IsolatedExecutionState.context
    end
  else
    raise ActiveRecordError, "Cannot steal connection, it is not currently leased."
  end
end

#throw_away!

Removes the connection from the pool and disconnect it.

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 817

def throw_away!
  pool.remove self
  disconnect!
end

#translate_exception(exception, message:, sql:, binds:) (private)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 1318

def translate_exception(exception, message:, sql:, binds:)
  # override in derived class
  case exception
  when RuntimeError, ActiveRecord::ActiveRecordError
    exception
  else
    ActiveRecord::StatementInvalid.new(message, sql: sql, binds: binds, connection_pool: @pool)
  end
end

#translate_exception_class(native_error, sql, binds) (private)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 1267

def translate_exception_class(native_error, sql, binds)
  return native_error if native_error.is_a?(ActiveRecordError)

  message = "#{native_error.class.name}: #{native_error.message}"

  active_record_error = translate_exception(
    native_error, message: message, sql: sql, binds: binds
  )
  return active_record_error if active_record_error.equal?(native_error)

  active_record_error.set_backtrace(native_error.backtrace)

  begin
    raise active_record_error, cause: native_error
  rescue => error
    error
  end
end

#type_map (private)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 1257

def type_map
  if key = extended_type_map_key
    self.class::EXTENDED_TYPE_MAPS.compute_if_absent(key) do
      self.class.extended_type_map(**key)
    end
  else
    self.class::TYPE_MAP
  end
end

#unprepared_statement

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 408

def unprepared_statement
  cache = prepared_statements_disabled_cache.add?(object_id) if @prepared_statements
  yield
ensure
  cache&.delete(object_id)
end

#valid_raw_connection (private)

Similar to any_raw_connection, but ensures it is validated and connected. Any method called on this result still needs to be independently thread-safe, so it probably shouldn't talk to the server... but some drivers fail if they know the connection has gone away.

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 1243

def valid_raw_connection
  (@verified && @raw_connection) ||
    # `allow_retry: false`, to force verification: the block won't
    # raise, so a retry wouldn't help us get the valid connection we
    # need.
    with_raw_connection(allow_retry: false, materialize_transactions: false) { |conn| conn }
end

#valid_type?(type) ⇒ Boolean

This method is for internal use only.
[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 297

def valid_type?(type) # :nodoc:
  self.class.valid_type?(type)
end

#verified! (private)

Mark the connection as verified. Call this inside a #with_raw_connection block only when the block is guaranteed to exercise the raw connection.

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 1161

def verified!
  @last_activity = Process.clock_gettime(Process::CLOCK_MONOTONIC)
  @verified = true
end

#verify

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 840

def verify
  return if @verified
  return if (last_activity = seconds_since_last_activity) && last_activity < verify_timeout
  verify!
end

#verify!

Checks whether the connection to the database is still active (i.e. not stale). This is done under the hood by calling #active?. If the connection is no longer active, then this method will reconnect to the database.

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 849

def verify!
  if @needs_reconnect || !active?
    @lock.synchronize do
      if @unconfigured_connection
        attempt_configure_connection do
          @raw_connection = @unconfigured_connection
          @unconfigured_connection = nil
          @needs_reconnect = false
          configure_connection
          @last_activity = Process.clock_gettime(Process::CLOCK_MONOTONIC)
          @verified = true
          @allow_preconnect = true
        end
        return
      end

      reconnect!(restore_transactions: true)
    end
  end

  @last_activity = Process.clock_gettime(Process::CLOCK_MONOTONIC)
  @verified = true
end

#verify_timeout

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 238

def verify_timeout
  (@config[:verify_timeout] || 2).to_i
end

#warning_ignored?(warning) ⇒ Boolean (private)

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 1382

def warning_ignored?(warning)
  ActiveRecord.db_warnings_ignore.any? do |warning_matcher|
    warning.message.match?(warning_matcher) || warning.code.to_s.match?(warning_matcher)
  end
end

#with_raw_connection(allow_retry: false, materialize_transactions: true) (private)

Lock the monitor, ensure we're properly connected and transactions are materialized, and then yield the underlying raw connection object.

If allow_retry is true, a connection-related exception will cause an automatic reconnect and re-run of the block, up to the connection's configured #connection_retries setting and the configured #retry_deadline limit. (Note that when allow_retry is true, it's possible to return without having marked the connection as verified. If the block is guaranteed to exercise the connection, consider calling #verified! to avoid needless verification queries in subsequent calls.)

If materialize_transactions is false, the block will be run without ensuring virtual transactions have been materialized in the DB server's state. The active transaction will also remain clean (if it is not already dirty), meaning it's able to be restored by reconnecting and opening an equivalent-depth set of new transactions. This should only be used by transaction control methods, and internal transaction-agnostic queries.

It's not the primary use case, so not something to optimize for, but note that this method does need to be re-entrant: materialize_transactions will re-enter if it has work to do, and the yield block can also do so under some circumstances.

In the latter case, we really ought to guarantee the inner call will not reconnect (which would interfere with the still-yielded connection in the outer block), but we currently provide no special enforcement there.

[ GitHub ]

  
# File 'activerecord/lib/active_record/connection_adapters/abstract_adapter.rb', line 1124

def with_raw_connection(allow_retry: false, materialize_transactions: true)
  @lock.synchronize do
    reconnectable = ensure_connection_ready(allow_retry:, materialize_transactions:)

    budget = build_retry_budget(allow_retry:, reconnectable:)

    begin
      yield @raw_connection
    rescue => original_exception
      translated_exception = translate_exception_class(original_exception, nil, nil)
      invalidate_transaction(translated_exception)
      downgrade_connection_after_error(translated_exception)

      if attempt_retry(translated_exception, budget)
        if retryable_connection_error?(translated_exception)
          ensure_connection_ready(allow_retry:, materialize_transactions: false)
        end
        retry
      end

      raise translated_exception
    rescue Exception
      # A non-StandardError (a Timeout, or a fiber scheduler's cancel) abandoned
      # the query partway through, so we mark the connection unverified, just as
      # a failed query would, forcing a reconnect before it's used again.
      @last_activity = nil
      @verified = false
      raise
    ensure
      dirty_current_transaction if materialize_transactions
    end
  end
end