Module Sequel::Postgres
In: lib/sequel/adapters/postgres.rb
lib/sequel/adapters/shared/postgres.rb

Top level module for holding all PostgreSQL-related modules and classes for Sequel. There are a few module level accessors that are added via metaprogramming. These are:

  • client_min_messages (only available when using the native adapter) - Change the minimum level of messages that PostgreSQL will send to the the client. The PostgreSQL default is NOTICE, the Sequel default is WARNING. Set to nil to not change the server default.
  • force_standard_strings - Set to false to not force the use of standard strings
  • use_iso_date_format (only available when using the native adapter) - Set to false to not change the date format to ISO. This disables one of Sequel‘s optimizations.

Changes in these settings only affect future connections. To make sure that they are applied, they should generally be called right after the Database object is instantiated and before a connection is actually made. For example, to use whatever the server defaults are:

  DB = Sequel.postgres(...)
  Sequel::Postgres.client_min_messages = nil
  Sequel::Postgres.force_standard_strings = false
  Sequel::Postgres.use_iso_date_format = false
  # A connection to the server is not made until here
  DB[:t].all

The reason they can‘t be done earlier is that the Sequel::Postgres module is not loaded until a Database object which uses PostgreSQL is created.

Classes and Modules

Module Sequel::Postgres::AdapterMethods
Class Sequel::Postgres::Adapter
Class Sequel::Postgres::Database
Class Sequel::Postgres::Dataset

Constants

PG_TYPES = {}   Hash with integer keys and proc values for converting PostgreSQL types.
PG_TYPE_PROCS = { [16] => lambda{|s| s == 't'}, # boolean [17] => lambda{|s| ::Sequel::SQL::Blob.new(Adapter.unescape_bytea(s))}, # bytea [20, 21, 22, 23, 26] => lambda{|s| s.to_i}, # integer [700, 701] => lambda{|s| s.to_f}, # float [790, 1700] => lambda{|s| BigDecimal.new(s)}, # numeric [1082] => lambda{|s| @use_iso_date_format ? Date.new(*s.split("-").map{|x| x.to_i}) : Sequel.string_to_date(s)}, # date [1083, 1266] => lambda{|s| Sequel.string_to_time(s)}, # time [1114, 1184] => lambda{|s| Sequel.string_to_datetime(s)}, # timestamp }   Use a single proc for each type to conserve memory
CONVERTED_EXCEPTIONS = []   Array of exceptions that need to be converted. JDBC uses NativeExceptions, the native adapter uses PGError.
SELECT_PK = proc do |schema, table| <<-end_sql SELECT pg_attribute.attname FROM pg_class, pg_attribute, pg_index, pg_namespace WHERE pg_class.oid = pg_attribute.attrelid AND pg_class.relnamespace = pg_namespace.oid AND pg_class.oid = pg_index.indrelid AND pg_index.indkey[0] = pg_attribute.attnum AND pg_index.indisprimary = 't' #{"AND pg_namespace.nspname = '#{schema}'" if schema} AND pg_class.relname = '#{table}' end_sql

Attributes

client_min_messages  [RW]  By default, Sequel sets the minimum level of log messages sent to the client to WARNING, where PostgreSQL uses a default of NOTICE. This is to avoid a lot of mostly useless messages when running migrations, such as a couple of lines for every serial primary key field.
force_standard_strings  [RW]  By default, Sequel forces the use of standard strings, so that ’\’ is interpreted as \ and not \. While PostgreSQL defaults to interpreting plain strings as extended strings, this will change in a future version of PostgreSQL. Sequel assumes that SQL standard strings will be used.
use_iso_date_format  [RW]  As an optimization, Sequel sets the date style to ISO, so that PostgreSQL provides the date in a known format that Sequel can parse faster. This can be turned off if you require a date style other than ISO.

[Validate]