sequel 5.20.0 → 5.49.0

data/doc/release_notes/4.31.0.txt

@@ -1,57 +0,0 @@
-= Improvements
-
-* Sequel now works with ruby 2.3's --enable-frozen-string-literal,
-  and all of the library files are set to use frozen string
-  literals by default.
-
-  A couple adapters and extensions depend on libraries that have
-  issues with frozen string literals.  Pull requests have been sent
-  to each of those dependencies.
-
-* The migrators will now raise an exception if a migration file
-  contains no migrations or more than one migration.
-
-* The jdbc/postgresql adapter now supports using PostgreSQL specific
-  types in bound variables.  Note that the current version of
-  jdbc-postgres (9.4.1204) has regressions that affect this, users
-  who need this support should stick with jdbc-postgres 9.4.1200 or
-  below.
-
-* The jdbc/postgresql adapter now works around a regression in Java
-  method lookup in JRuby 9.0.5.0
-
-* The setter methods added by the association_pks plugin now do type
-  casting immediately, instead of right before the data will be used.
-  This makes them more similar to column setter methods, and ensures
-  that future calls to the getters that use cached values will
-  return correctly typecast data.
-
-* The PostgreSQL array parser in the pg_array extension now handles
-  arrays with explicit bounds.  The explicit bounds are ignored, so
-  such values do not round trip, and there is currently no support for
-  creating arrays with explicit bounds.
-
-* Creating a table with a simple non-incrementing primary key and a
-  self-referential foreign key now works correctly on MySQL:
-
-    DB.create_table!(:table) do
-      Integer :id, :primary_key=>true
-      foreign_key :fk, :table
-    end
-
-* Database#disconnect in the oracle adapter now works correctly on
-  more recent versions of oci8 where #logoff can raise OCIException
-  instead of OCIInvalidHandle.
-
-= Backwards Compatibility
-
-* The pg_array extension no longer defines
-  Sequel::Postgres::PGArray::JSONCreator.  This should only affect
-  backwards compatibility if you were accessing the constant directly.
-  The :parser option to Sequel::Postgres::PGArray.register is also no
-  longer respected, but that should not affect backwards compatibility.
-
-* The Sequel::Model#convert_cpk_array private method that was added by
-  the association_pks plugin has been removed.
-  Sequel::Model#convert_pk_array handles both simple and composite
-  primary keys now.

data/doc/release_notes/4.32.0.txt

@@ -1,132 +0,0 @@
-= New Features
-
-* A no_auto_literal_strings extension has been added, which removes the
-  automatic usage of strings in filter arguments as literal SQL code.
-  By default, if you do:
-
-    DB[:albums].where("name > 'N'")
-
-  By default Sequel will treat "name > 'N'" as SQL code.  However,
-  this makes it much easier to introduce SQL injection:
-
-    # SQL Injection vulnerability in default Sequel
-    DB[:albums].where("name > 'params[:letter]'")
-
-  Sequel does support using placeholders when using literal strings:
-
-    # Safe in default Sequel
-    DB[:albums].where("name > ?", params[:letter])
-
-  However, if you forget to use placeholders, you can end up with SQL
-  injection.  Accidental usage of filter strings derived from user
-  input as literal SQL code is probably the most common SQL injection
-  vector in applications using Sequel.
-
-  With the no_auto_literal_strings extension, passing a plain string
-  as the first or only argument to a filter method raises an
-  exception.  If you want to use literal SQL code, you have to do so
-  explicitly:
-
-    DB[:albums].where(Sequel.lit("name > 'N'"))
-
-  You can also specify placeholders when using Sequel.lit:
-
-    DB[:albums].where(Sequel.lit("name > ?", params[:letter]))
-
-  Note that in many cases, you can avoid using literal SQL strings
-  completely:
-
-    DB[:albums].where{|v| v.name > params[:letter]}
-
-* one_through_one associations now support a setter method:
-
-    Foo.one_through_one :bar
-
-    foo = Foo[1]
-    foo.bar = Bar[2]
-    foo.bar = nil
-
-  This will check the current entry in the join table, and based on
-  the argument and the current entry, run a DELETE, INSERT, or UPDATE
-  query, or take no action if the join table is already in the
-  correct state.
-
-* Model.default_association_options has been added, which supports
-  default options for all future associations.  You can use this to
-  do:
-
-    Model.default_association_options = {:read_only=>true}
-
-  Which makes associations not create modification methods by default.
-  You could still create the modification methods by passing
-  :read_only=>true when creating association.
-
-* The tactical_eager_loading plugin now supports two additional
-  options when calling an association method: :eager and
-  :eager_reload. Example:
-
-    artist = Artist.all.first
-
-    # Loads all albums for all of the artists,
-    # and all tracks for all of those albums
-    artist.albums(:eager=>:tracks)
-
-    # Reload the artists association for all artists
-    artist.albums(:eager_reload=>true)
-
-  You can also use the :eager option for an eager loading callback:
-
-    # Eagerly load the albums with names starting with A-M
-    artist.albums(:eager=>proc{|ds| ds.where(:name > 'N')})
-
-* The association_pks plugin now supports an :association_pks_nil
-  association option in the association_pks setter, for determining
-  how nil values should be handled.
-
-  In Sequel <4.31.0, if you provided nil, it would either raise an
-  exception immediately if :delay_pks was not set, or on saving if
-  :delay_pks was set.
-
-  In Sequel 4.31.0, if :delay_pks was not set, it would remove all
-  associated rows.  If :delay_pks was set, it would do nothing.
-
-  You can now set :association_pks_nil=>:remove to remove all
-  associated values on nil, or :association_pks_nil=>:ignore to ignore
-  a nil value passed to the method.  Without :association_pks_nil set,
-  an exception will be raised.
-
-* Dataset#delete_from has been added on MySQL, allowing deletions from
-  multiple tables in a single query:
-
-    DB[:a].join(:b, :a_id=>:id).delete_from(:a, :b).delete
-    # DELETE a, b FROM a INNER JOIN b ON (b.a_id = a.id)
-
-* The JDBC schema parser now includes a :remarks entry for each
-  column, which contains comments on the column.
-
-= Other Improvements
-
-* The setter method added by the association_pks plugin now handles
-  the empty array correctly when :delay_pks is set.  Previously, if
-  the empty array was passed, Sequel made no modifications in this
-  case.  Sequel now correctly removes all associated values if an
-  empty array is passed.
-
-* The eager_each plugin now handles eager loading when using
-  Dataset#first and related methods.  Previously, the behavior was
-  unspecified. In Sequel <4.27.0 Dataset#first did eager loading
-  correctly in the eager case, but incorrectly in the eager_graph
-  case.  In Sequel 4.27.0-4.31.0, it did not do eager loading in
-  either case.
-
-* The tactical_eager_loading plugin will not automatically eager load
-  if passing a proc or block to an association method, since the proc
-  or block could be specific to the receiver.
-
-* Sequel now uses a mutex to synchronize access to the association
-  cache on MRI, as it does on other ruby implementations.
-
-= Backwards Compatibility
-
-* See above for changes in eager_each and association_pks plugin
-  behavior.

data/doc/release_notes/4.33.0.txt

@@ -1,88 +0,0 @@
-= New Features
-
-* A Sequel::Model.require_valid_table accessor has been added. This
-  setting is false for backwards compatibility, but if set to true,
-  will raise an error you try to create a model class where an
-  invalid table name is used or the schema or columns cannot be
-  determined.  This makes it easier to catch bugs, as things will
-  fail fast, but it means that you must change code like:
-
-    class Foo < Sequel::Model
-      set_dataset :my_foos
-    end
-
-  to:
-
-    class Foo < Sequel::Model(:my_foos)
-    end
-
-  as otherwise Foo will attempt to use the foos table by default
-  when creating the class, which will raise an error as it is not
-  the correct table name.
-
-* Sequel::Database#transaction now supports a :savepoint=>:only
-  option, which will create a savepoint if already inside a
-  transaction, but will yield without creating a transaction if
-  not inside a transaction.  The use case for this is when you
-  are running code that may raise an exception, and you don't
-  want to invalidate the current transaction state.
-
-= Other Improvements
-
-* The graph_each extension now splits results into subhashes when
-  using Sequel::Dataset#first, as it did before Sequel 4.27.0.
-
-* On PostgreSQL, Dataset#insert_conflict now accepts an array of
-  columns as the value for the :target option.
-
-* You can now pass a Sequel::SQL::Identifier or a
-  Sequel::SQL::QualifiedIdentifer as the table argument when creating
-  a foreign key.  Previously, only symbols were supported, and using
-  other values required specifying the :table option.  So this will
-  now work to reference a table that includes a double underscore:
-
-    foreign_key :foo_id, Sequel.identifier(:fo__oo)
-
-* Creating model classes inside a transaction on PostgreSQL where
-  the implicit table name isn't correct no longer causes the
-  transaction to fail.
-
-  Similar issues were also fixed in the boolean_readers,
-  boolean_subsets, and class_table_inheritance plugins.
-
-* On PostgreSQL, You can now use the :qualify=>true option in the
-  schema dumper, to dump using schema-qualified table names.
-
-* On Microsoft SQL Server, the set_column_allow_null and
-  set_column_not_null alter table methods now work on varchar(max),
-  text, and similar columns.
-
-* On Oracle, Sequel::Database#sequence_for_table now returns nil if
-  given a table that doesn't exist or that the user does not have
-  access to.
-
-* Passing arbitrary objects to a model association method now
-  indicates that the association should be reloaded, which was
-  used to work but was broken in Sequel 4.32.0.
-
-* It is now possible to raise Sequel::ValidationFailed and
-  Sequel::HookFailed without an argument.
-
-= Backwards Compatibility
-
-* Sequel::Model no longer swallows many errors when subclassing or
-  setting datasets.  While this should hopefully not affect backwards
-  compatibility, it may break things where the methods were raising
-  exceptions.  If this does break backwards compatibility, it is
-  most likely because it is no longer hiding another bug that should
-  be fixed.  Specific changes include:
-
-  * Model.inherited no longer rescues exceptions raised by set_dataset
-  * When subclassing a model that has a dataset, the columns and
-    schema are just copied from the superclass
-  * Only Sequel::Error is rescued in calls to columns and schema,
-    before it would rescue StandardError.
-
-* The Sequel.firebird and Sequel.informix adapter methods have been
-  removed, they are no longer needed as the firebird and informix
-  adapters were removed a few versions back.

data/doc/release_notes/4.34.0.txt

@@ -1,86 +0,0 @@
-= New Features
-
-* A duplicate_columns_handler extension has been added, for printing a
-  warning or raising an exception if a dataset returns multiple
-  columns with the same name.  You can set this globally for the
-  Database:
-
-    DB.extension :duplicate_columns_handler
-    DB.opts[:on_duplicate_columns] = :warn
-    DB.opts[:on_duplicate_columns] = proc do |columns|
-      columns.include?(:foo) ? :raise : :ignore
-    end
-
-  or for specific datasets:
-
-    ds = DB[:table].extension(:duplicate_columns_handler)
-    ds = ds.on_duplicate_columns(:raise)
-    ds = ds.on_duplicate_columns do |columns|
-      columns.include?(:foo) ? :raise : :ignore
-    end
-
-  This makes it easier to detect when duplicate columns are returned,
-  which in some cases can cause undesired behavior, such as the values
-  for later columns of the same name overwriting values for earlier
-  columns.
-
-* The Dataset#to_hash, #to_hash_groups, #select_hash, and
-  #select_hash_groups methods now take an options hash as a third
-  argument.  This options hash can now contain a :hash option, which
-  specifies the object in which the resulting values should be
-  placed.  You can use this to have the values inserted into a
-  custom hash, or another object responding to #[] and #[]=.
-
-* A validators_operator validation has been added to the
-  validation_helpers plugin:
-
-    class Example < Sequel::Model
-      def validate
-        super
-        validates_operator(:>, 3, :column1)
-        validates_operator(:<=, 4, [:column2, :column3])
-      end
-    end
-
-* The pg_range extension now adds a #register_range_type Database
-  method, supporting per-Database custom range types:
-
-    DB.register_range_type('timerange')
-
-* The dataset_associations plugin now supports a
-  :dataset_associations_join association option on associations that
-  use joined datasets.  This option will have the datasets returned
-  by the dataset association methods join to the same tables that
-  would be joined when retriving the associated objects, allowing
-  selected columns, orders, and filters that reference columns in
-  the joined tables to work correctly.
-
-* The Database :preconnect option can now be set to :concurrently,
-  which will create the connections in separate threads. This can
-  significantly speed up preconnection in high-latency environments.
-
-* The Database :name option is now supported, holding an arbitrary
-  name for the database.  Currently, it is only used in PoolTimeout
-  exception messages, but it may be used in other places in the
-  future.
-
-= Other Improvements
-
-* The prepared_statements_safe plugin now works correctly when using
-  CURRENT_DATE and CURRENT_TIMESTAMP default values for columns.
-
-* Sequel now recognizes an addition unique constraint violation on
-  Microsoft SQL Server.
-
-* PoolTimeout exception messages now include the server/shard to which
-  the connection was attempted when using the sharded threaded
-  connection pool.
-
-= Backwards Compatibility
-
-* Users of sequel_pg should upgrade to 1.6.17, as older versions of
-  sequel_pg may not work with Sequel 4.34.0+.
-
-* Any custom extensions that override Dataset#to_hash,
-  #to_hash_groups, #select_hash, and #select_hash_groups need to
-  be modified to add support for accepting the options hash.

data/doc/release_notes/4.35.0.txt

@@ -1,130 +0,0 @@
-= Forwards Compatibility
-
-* Ruby 2.4 will unify the Fixnum and Bignum classes into the Integer
-  class, making both Fixnum and Bignum references to Integer. This
-  will have the affect of changing the behavior of Sequel migrations
-  that use a reference to the Bignum class.  
-
-  For example, code like this will change behavior in ruby 2.4:
-
-    DB.create_table(:table) do
-      add_column :column, Bignum
-    end
-    # or:
-    DB.get(Sequel.cast('1', Bignum))
-
-  as this references the Bignum class. On ruby <2.4, this will create
-  a 64-bit integer column, on ruby 2.4+, it will create a 32-bit
-  integer column.
-
-  Code like this will be fine and does not need changing:
-
-    DB.create_table(:table) do
-      Bignum :column
-    end
-
-  as this calls the Bignum method.
-
-  Sequel now supports the :Bignum symbol as a generic type, so you
-  can now switch references to the Bignum class to the :Bignum
-  symbol whenever you want a generic 64-bit integer type:
-
-    DB.create_table(:table) do
-      add_column :column, :Bignum
-    end
-    # or:
-    DB.get(Sequel.cast('1', :Bignum))
-
-  Note that you should only do this if you are using Sequel 4.35.0+,
-  as previous versions of Sequel will treat the :Bignum symbol as
-  a database-specific type named Bignum.
-
-= New Features
-
-* A Sequel::Database#log_connection_info accessor has been added.  If
-  set to true, this includes connection information in Sequel's query
-  log.  In threaded connection pools (the default), this makes it
-  simple to see which connection is executing which queries.
-
-    DB.log_connection_info = true
-    DB.get(1)
-    # Logged: (0.000004s) (conn: 9713390226040) SELECT 1 AS v LIMIT 
-
-* Sequel::Model#lock! now supports an optional lock style, instead
-  of always using FOR UPDATE (which is still the default):
-
-    Example.first.lock!('FOR NO KEY UPDATE')
-    #=> SELECT * FROM examples WHERE id = 1 FOR NO KEY UPDATE LIMIT 1
-
-* Sequel::Dataset#skip_locked has been added, which skips locked rows
-  when returning query results. This is useful whenever you are
-  implementing a queue or similar data structure.  Currently, this is
-  supported on PostgreSQL 9.5+, Oracle, and Microsoft SQL Server.
-
-* An sql_comments extension has been added for setting SQL comments
-  on queries:
-
-    ds = DB[:table].comment("Some Comment").all
-    # SELECT * FROM table -- Some Comment
-    #
-
-  All consecutive whitespace in the comment is replaced by a
-  single space, and the comment ends in a newline so that it works
-  correctly in subqueries.
-
-  This extension is mostly useful if you are doing analysis of your
-  database server query log and want to include higher level
-  information about the query in the comment.
-
-* A server_logging extension has been added, which includes
-  server/shard information in the query log, if connection info
-  is being logged.
-
-    DB.extension :server_logging
-    DB.log_connection_info = true
-    DB.get(1)
-    # Logged: (0.000004s) (conn: 9712828677240, server: read_only)
-    #         SELECT 1 AS v LIMIT 1 
-    DB[:a].insert(:b=>1)
-    # Logged: (0.000003s) (conn: 9712534040260, server: default)
-    #         INSERT INTO a (b) VALUES (1)
-
-* On PostgreSQL, Database#full_text_search now supports a
-  :headline option for adding an extract of the matched text to
-  the SELECT list.
-
-* Sequel::Postgres::PGRange#cover? has been added to the pg_range
-  extension, which works with empty, unbounded, and exclusive
-  beginning ranges.  Previously, using #cover? with these ranges
-  would raise an exception.  Note that cover? is now always
-  defined, where previously it was only defined on ruby 1.9+.
-
-= Other Improvements
-
-* The jdbc adapters now work correctly on JRuby 9.1.  Previously,
-  some parts were broken on JRuby 9.1 due to frozen string literal
-  issues.
-
-* Sequel::Dataset#to_hash and #to_hash_groups now work correctly for
-  model datasets doing eager loading.
-
-* Using Sequel::Database#transaction with the :rollback=>:always
-  option now automatically uses a savepoint if supported when run
-  inside another transaction.  If savepoints are not supported,
-  using :rollback=>:always inside a transaction will now raise an
-  exception.
-
-* The delay_add_association plugin now handles hashes and primary keys
-  passed to the add_* association methods.
-
-* The json_serializer :include option now works correctly when using
-  *_to_many associations with the association_proxies plugin.
-
-* The schema_dumper extension now recognizes bool as a boolean type,
-  for consistency with the Database schema parser.
-
-= Backwards Compatibility
-
-* Custom adapters should switch from using log_yield to
-  log_connection_yield so that they work correctly when using
-  log_connection_info.

data/doc/release_notes/4.36.0.txt

@@ -1,116 +0,0 @@
-= New Features
-
-* Sequel::Model::Model() has been added, which allows for
-  Sequel::Model() like behavior where the base class used is a
-  subclass of Sequel::Model.  To make it easier to use,
-  Sequel::Model.def_Model has also been added, which takes a module
-  and adds a Model() method to the module that calls Model() on the
-  receiver.
-
-  A :class_namespace association option has been added to make it
-  possible to set a default namespace for the :class option if given
-  as a symbol or string.
-
-  Sequel::Model.Model.cache_anonymous_models has been added and
-  controls whether to cache anonymous model subclasses created by
-  Sequel::Model::Model() on a per-class basis.
-
-  These changes are designed to make it easier to use namespaced
-  models, for example:
-
-    module Foo
-       Model = Class.new(Sequel::Model)
-       Model.def_Model(self)
-       DB = Model.db = Sequel.connect(ENV['FOO_DATABASE_URL'])
-       Model.plugin :prepared_statements
-       Model.default_association_options[:class_namespace] = 'Foo'
-
-       class Bar < Model
-         # Uses Foo::DB[:bars] as dataset
-
-         # Implicitly uses Foo::Baz as associated class
-         one_to_many :bazes
-
-         # Uses Foo::Baz due to :class_namespace option
-         one_to_many :oldest_bazes, :class=>:Baz, :order=>:id
-       end
-
-       class Baz < Model(:my_baz)
-         # Uses Foo::DB[:my_baz] as dataset
-
-         # Implicitly uses Foo::Bar as associated class
-         one_to_many :bars
-
-         # Uses Foo::Bar due to :class_namespace option
-         one_to_many :oldest_bars, :class=>:Bar, :order=>:id
-       end
-     end
-
-* A string_agg extension has been added for aggregate string
-  concatentation support on PostgreSQL 9+, SQLAnywhere 12+,
-  Oracle11g+, DB 9.7+, MySQL, HSQLDB, H2, and CUBRID:
-
-    DB.extension :string_agg
-    ds = DB[:table]
-    ds.get(Sequel.string_agg(:c))             # ',' default separator
-    ds.get(Sequel.string_agg(:c, ' - '))      # custom separator
-    ds.get(Sequel.string_agg(:c).order(:bar)) # force order
-    ds.get(Sequel.string_agg(:c).distinct)    # remove duplicates
-
-* A connection_expiration extension has been added, for automatically
-  removing connections from the connection pool after they have been
-  open for a given amount of time (4 hours by default).
-
-* Support for <, <=, >, and >= operator validations when using integer
-  and string arguments has been added to the constraint_validations
-  extension and plugin.
-
-* Sequel::SQL::Function#order has been added to support ordered
-  aggregate functions:
-
-    Sequel.function(:foo, :bar).order(:baz)
-    # foo(bar ORDER BY baz)
-
-= Other Improvements
-
-* The validates_operator validation in validation_helpers now
-  considers nil values as invalid unless :allow_nil or a similar
-  option is used.  Previously, using validates_operator with a nil
-  value would probably raise a NoMethodError.  This makes
-  validates_operator more similar to other validations.
-
-* The threaded connection pools no longer hold the pool mutex when
-  disconnecting connections, which is useful if the driver blocks
-  when disconnecting connections.
-
-* The connection_validator extension no longer holds a reference
-  to connections that have been disconnected.
-
-* The connection_validator extension no longer overwrites the
-  connection_validation_timeout if loaded a second time.
-
-* Sequel now closes cursors as soon as it is done using them in the
-  oracle adapter, instead of waiting for GC to clean them up.
-
-* Sequel now handles disconnect errors that occur when literalizing
-  strings in the mysql2 and postgres adapters.
-
-= Backwards Compatibility
-
-* Using the Bignum class as a generic type is now deprecated.  As
-  announced in the 4.35.0 release notes, ruby 2.4 is unifying the
-  Fixnum and Bignum classes into Integer, which results in the
-  behavior of the Bignum class changing.  4.35.0 added support for
-  using the :Bignum symbol as a generic 64-bit integer type, and
-  Sequel users now need to switch to that to avoid the deprecation
-  warning.
-
-  Sequel 4.41.0 (to be released in December), will drop support
-  for using the Bignum class as a generic type.  This is being done
-  before the release of ruby 2.4 to hopefully make it unlikely that
-  users will be subject to a behavior changes when upgrading ruby
-  versions.
-
-  Related to this change, external adapters need to switch from
-  overriding Database#type_literal_generic_bignum to 
-  Database#type_literal_generic_bignum_symbol.

data/doc/release_notes/4.37.0.txt

@@ -1,50 +0,0 @@
-= New Features
-
-* Database#values has been added on SQLite#3.8.3+, operating similarly
-  to the support on PostgreSQL:
-
-    DB.values([[1, 2], [3, 4]]).select_map([:column1, :column2])
-    # => [[1, 2], [3, 4]]
-
-* Regular expressions in dataset filters are now supported on Oracle
-  10g+:
-
-    DB[:t].where(:c=>/re/)
-    # SELECT * FROM "T" WHERE REGEXP_LIKE("C",'re')
-
-= Other Improvements
-
-* Sequel now supports the use of native prepared statements and bound
-  variables in the mysql2 adapter, when mysql2 0.4+ is used.
-  Previously, the mysql2 adapter supported database prepared
-  statements, but variables were always literalized.  That is still
-  supported when mysql2 <0.4 is used.
-
-* The connection pool now removes connections if it detects a
-  disconnect error that is not raised as a
-  Sequel::DatabaseDisconnectError.  Such exceptions are reraised
-  without converted them to Sequel::DatabaseDisconnectError, but the
-  related connection is now removed from the pool.
-
-* The reversible migration support now handles add_constraint with an
-  options hash as the first argument.
-
-* ASTTransformer now handles Sequel.extract, allowing Dataset#qualify
-  and other uses of ASTTransformer to work with such values.
-
-* The create_view :columns option is now suppported on SQLite 3.9.0+.
-
-* An additional disconnect error is now recognized in the postgres
-  adapter.
-
-* A frozen string literal issue has been fixed when multiple different 
-  database connection approaches have failed in the jdbc adapter.
-
-= Backwards Compatibility
-
-* External database adapters need to make sure that
-  Database#database_error_classes returns a valid result if called
-  during Database#initialize.  If you have an external adapter where
-  one of the error classes depends on an argument given when
-  connecting (such as the connection string), you may have to make
-  some changes.