=== 0.3 (2007-10-20)

* Added stock transforms to Dataset#transform. Refactored Model.serialize.

* Added Database#logger= method for setting the database logger object.

* Fixed Model.[] to act as shortcut to Model.find when a hash is given (#71).

* Added support for old and new decimal types in MySQL adapter, and updated MYSQL_TYPES with MySQL 5.0 constants (#72).

* Implemented Database#disconnect method for all adapters.

* Fixed small bug in ArrayKeys module.

* Implemented model caching by primary key.

* Separated Model.find and Model.[] functionality. Model.find takes a filter. Model.[] is strictly for finding by primary keys.

* Enhanced Dataset#first to accept a filter block. Model#find can also now accept a filter block.

* Changed Database#[] to act as shortcut to #fetch if a string is given.

* Renamed Database#each to #fetch. If no block is given, the method returns an enumerator.

* Changed Dataset#join methods to correctly literalize values in join conditions (#70).

* Fixed #filter with ranges to correctly literalize field names (#69).

* Implemented Database#each method for quickly retrieving records with arbitrary SQL (thanks Aman Gupta).

* Fixed bug in postgres adapter where a LiteralString would be literalized as a regular String.

* Fixed SQLite insert with subquery (#68).

* Reverted back to hashes as default mode. Added Sequel.use_array_tuples and Sequel.use_hash_tuples methods.

* Fixed problem with arrays with keys when using #delete.

* Implemented ArrayKeys as substitute for ArrayFields.

* Added Dataset#each_hash method.

* Rewrote SQLite::Database#transaction to use sqlite3-ruby library implementation of transactions.

* Fixed Model.destroy_all to work correctly in cases where no before_destroy hook is defined and an after_destroy hook is defined.

* Restored Model.has_hooks? implementation.

* Changed Database#<< to strip comments and whitespace only when an array is given.

* Changed Schema::Generator#primary_key to accept calls with the type argument omitted.

* Hooks can now be prepended or appended by choice.

* Changed Model.subset to define filter method on the underlying dataset instead of the model class.

* Fixed Dataset#transform to work with array fields.

* Added Dataset#to_csv method.

* PrettyTable can now extract column names from arrayfields.

* Converted ado, dbi, odbc adapters to use arrayfields instead of hashes.

* Fixed composite key support.

* Fixed Dataset#insert_sql, update_sql to support array fields.

* Converted sqlite, mysql, postgres adapters to use arrayfields instead of hashes.

* Extended Dataset#from to auto alias sub-queries.

* Extended Dataset#from to accept hash for aliasing tables.

* Added before_update, after_update hooks.

=== 0.2.1.1 (2007-10-07)

* Added Date literalization to sqlite adapter (#60).

* Changed Model.serialize to allow calling it after the class is defined (#59).

* Fixed after_create hooks to allow calling save inside the hook (#58).

* Fixed MySQL quoting of sql functions (#57).

* Implemented rollback! global method for cancelling transactions in progress.

* Fixed =~ operator in Sequelizer.

* Fixed ODBC::Dataset#fetch_rows (thanks Dusty).

* Renamed Model.recreate_table to create_table!. recreate_table is deprecated and will issue a warning (#56).

=== 0.2.1 (2007-09-24)

* Added default implementation of Model.primary_key_hash.

* Fixed Sequel::Model() to set dataset for inherited classes.

* Rewrote Model.serialize to use Dataset#transform.

* Implemented Dataset#transform.

* Added gem spec for Windows (without ParseTree dependency).

* Added support for dynamic strings in Sequelizer (#49).

* Query branch merged into trunk.

* Implemented self-changing methods.

* Add support for ternary operator to Sequelizer.

* Fixed sequelizer to evaluate expressions if they don't involve symbols or literal strings.

* Added protection against using #each, #delete, #insert, #update inside query blocks.

* Improved Model#method_missing to deal with invalid attributes.

* Implemented Dataset#query.

* Added Dataset#group_by as alias for Dataset#group.

* Added Dataset#order_by as alias for Dataset#order.

* More model refactoring. Added support for composite keys.

* Added Dataset#empty? method (#46).

* Fixed Symbol#to_field_name to support names with numbers and upper-case characters (#45).

* Added install_no_doc rake task.

* Partial refactoring of model code.

* Refactored dataset-model association and added Dataset#set_row_filter method.

* Added support for case-sensitive regexps to mysql adapter.

* Changed mysql adapter to support encoding option as well.

* Added charset/encoding option to postgres adapter.

* Implemented Model.serialize (thanks Aman Gupta.)

* Changed Model.create to INSERT DEFAULT VALUES instead of (id) VALUES (null) (brings back #41.)

* Fixed Model.new to work without arguments.

* Added Model.no_primary_key method to allow models without primary keys.

* Added Model#this method (#42 thanks Duane Johnson).

* Fixed Dataset#insert_sql to use DEFAULT VALUES clause if argument is an empty hash.

* Fixed Model.create to work correctly when no argument is passed (#41).

=== 0.2.0.2 (2007-09-07)

* Dataset#insert can now accept subqueries.

* Changed Migrator.apply to return the version.

* Changed Sequel::Model() to cache intermediate classes so descendant classes can be reopened (#39).

* Added :charset option to MySQL adapter (#40).

* Fixed Dataset#exclude to add parens around NOT expression (#38).

* Fixed use of sub-queries with all comparison operators in block filters (#38).

* Fixed arithmetic expressions in block filters to not be literalized.

* Changed Symbol#method_missing to return LiteralString.

* Changed PrettyTable to right-align numbers.

* Fixed Model.create_table (thanks Duane Johnson.)

=== 0.2.0.1 (2007-09-04)

* Improved support for invoking methods with inline procs inside block filters.

=== 0.2.0 (2007-09-02)

* Fixed Model.drop_table (thanks Duane Johnson.)

* Dataset#each can now return rows for arbitrary SQL by specifying :sql option.

* Added spec for postgres adapter.

* Fixed Model.method_missing to work with new SQL generation.

* Fixed #compare_expr to support regexps.

* Fixed postgres, mysql adapters to support regexps.

* More specs for block filters. Updated README.

* Added support for globals and $X macros in block filters.

* Fixed Sequelizer to not fail if ParseTree or Ruby2Ruby gems are missing.

* Renamed String#expr into String#lit (#expr should be deprecated in future versions).

* Renamed Sequel::ExpressionString into LiteralString.

* Fixed Symbol#[] to return an ExpressionString, so as not to be literalized.

* Renamed Dataset::Expressions to Dataset::Sequelizer.

* Renamed Expressions#format_re_expression to match_expr.

* Renamed Expressions#format_eq_expression to compare_expr.

* Added support for Regexp in MySQL adapter.

* Refactored Regexp expressions into a separate #format_re_expression method.

* Added support for arithmetic in proc filters.

* Added support for nested proc expressions, more specs.

* Added support for SQL function using symbols, e.g. :sum[:x].

* Fixed deadlock bug in ConnectionPool.

* Removed deprecated old expressions.rb.

* Rewrote Proc filter feature using ParseTree.

* Added support for additional functions on columns using Symbol#method_missing.

* Added support for supplying filter block to DB#[] method, to allow stuff like DB[:nodes] {:path =~ /^icex1/}.

=== 0.1.9.12 (2007-08-26)

* Added spec for PrettyTable.

* Added specs for Schema::Generator and Model (#36 thanks technoweenie).

* Fixed Sequel::Model.set_schema (#36 thanks technoweenie.)

* Added support for no options on Schema::Generator#foreign_key (#36 thanks technoweenie.)

* Implemented (restored?) Schema::Generator#primary_key_name (#36 thanks technoweenie.)

* Better spec code coverage.

=== 0.1.9.11 (2007-08-24)

* Changed Dataset#set_model to allow supplying additional arguments to the model's initialize method (#35). Thanks Sunny Hirai.

=== 0.1.9.10 (2007-08-22)

* Changed schema generation code to generate separate statements for CREATE TABLE and each CREATE INDEX (#34).

* Refactored Dataset::SQL#field_name for better support of different field quoting standards by specific adapters.

* Added #current_page_record_count for paginated datasets.

* Removed Database#literal and included Dataset::SQL instead.

* Sequel::Dataset:SQL#field_name can now take a hash (as well as #select and any method that uses #field_name) for aliasing column names.  E.g. DB[:test].select(:_qqa => 'Date').sql #=> 'SELECT _qqa AS Date FROM test'.

* Moved SingleThreadedPool to lib/sequel/connection_pool.rb.

* Changed SQLite::Dataset to return affected rows for #delete and #update (#33).

* ADO adapter: Added use of Enumerable for Recordset#Fields, playing it safe and moving to the first row before getting results, and changing the auto_increment constant to work for MSSQL.

=== 0.1.9.9 (2007-08-18)

* New ADO adapter by cdcarter (#31).

* Added automatic column aliasing to #avg, #sum, #min and #max (#30).

* Fixed broken Sequel::DBI::Dataset#fetch_rows (#29 thanks cdcarter.)

=== 0.1.9.8 (2007-08-15)

* Fixed DBI adapter.

=== 0.1.9.7 (2007-08-15)

* Added support for executing batch statements in sqlite adapter.

* Changed #current_page_record_range to return 0..0 for an invalid page.

* Fixed joining of aliased tables.

* Improved Symbol#to_field_name to prevent false positives.

* Implemented Dataset#multi_insert with :commit_every option.

* More docs for Dataset#set_model.

* Implemented automatic creation of convenience methods for each adapter (e.g. Sequel.sqlite etc.)

=== 0.1.9.6 (2007-08-13)

* Refactored schema definition code. Gets rid of famous primary_key problem as well as other issues (e.g. issue #22).

* Added #pagination_record_count, #page_range and #current_page_record_range for paginated datasets.

* Changed MySQL adapter to automatically reconnect (issue #26).

* Changed Sequel() to accept variable arity.

* Added :elements option to column definition, in order to support ENUM and SET types.

=== 0.1.9.5 (2007-08-12)

* Fixed migration docs.

* Removed dependency on PGconn in Schema class.

=== 0.1.9.4 (2007-08-11)

* Added Sequel.dbi convenience method for using DBI connection strings to open DBI databases.

=== 0.1.9.3 (2007-08-10)

* Added support for specifying field size in schema definitions (thanks Florian Aßmann.)

* Added migration code based on work by Florian Aßmann.

* Reintroduced metaid dependency. No need to keep a local copy of it.

=== 0.1.9.2 (2007-07-24)

* Removed metaid dependency. Re-factored requires in lib/sequel.rb.

=== 0.1.9.1 (2007-07-22)

* Improved robustness of MySQL::Dataset#field_name.

* Added Sequel.single_threaded= convenience method.

=== 0.1.9 (2007-07-21)

* Fixed #update_sql and #insert_sql to support field quoting by calling #field_name.

* Implemented automatic data type conversion in mysql adapter.

* Added support for boolean literals in mysql adapter.

* Added support for ORDER and LIMIT clauses in UPDATE statements in mysql adapter.

* Implemented correct field quoting (using back-ticks) in mysql adapter.

* Wrote basic MySQL spec.

* Fixd MySQL::Dataset to return correct data types with symbols as hash keys.

* Removed discunctional MySQL::Database#transaction.

* Added support for single threaded operation.

* Fixed bug in Dataset#format_eq_expression where Range objects would not be literalized correctly.

* Added parens around postgres LIKE expressions using regexps.

=== 0.1.8 (2007-07-10)

* Implemented Dataset#columns for retrieving the columns in the result set.

* Updated Model with changes to how model-associated datasets work.

* Beefed-up specs. Coverage is now at 95.0%.

* Added support for polymorphic datasets.

* The adapter dataset interface was simplified and standardized. Only four methods need be overriden: fetch_rows, update, insert and delete.

* The Dataset class was refactored. The bulk of the dataset code was moved into separate modules.

* Renamed Dataset#hash_column to Dataset#to_hash.

* Added some common pragmas to sqlite adapter.

* Added Postgres::Dataset#analyze for EXPLAIN ANALYZE queries.

* Fixed broken Postgres::Dataset#explain.

=== 0.1.7

* Removed db.synchronize wrapping calls in sqlite adapter.

* Implemented Model.join method to restrict returned columns to the model table (thanks Pedro Gutierrez).

* Implemented Dataset#paginate method.

* Fixed after_destroy hook.

* Improved Dataset#first and #last to accept a filter hash.

* Added Dataset#[]= method.

* Added Sequel() convenience method.

* Fixed Dataset#first to include a LIMIT clause for a single record.

* Small fix to Postgres driver to return a primary_key value for the inserted record if it is specified in the insertion values (thanks Florian Aßmann and Pedro Gutierrez).

* Fixed Symbol#DESC to support qualified notation (thanks Pedro Gutierrez).

=== 0.1.6

* Fixed Model#method_missing to raise for an invalid attribute.

* Fixed PrettyTable to print model objects (thanks snok.)

* Fixed ODBC timestamp conversion to return DateTime rather than Time object (thanks snok.)

* Fixed Model.method_missing (thanks snok.)

* Model.method_missing now creates stubs for calling Model.dataset methods. Methods like Model.each etc are removed.

* Changed default join type to INNER JOIN (thanks snok.)

* Added support for literal expressions, e.g. DB[:items].filter(:col1 => 'col2 - 10'.expr).

* Added Dataset#and.

* SQLite adapter opens a memory DB if no database is specified, e.g. Sequel.open 'sqlite:/'.

* Added Dataset#or, pretty nifty.

=== 0.1.5

* Fixed Dataset#join to support multiple joins. Added #left_outer_join, #right_outer_join, #full_outer_join, #inner_join methods.

=== 0.1.4

* Added String#split_sql.

* Implemented Array#to_sql and String#to_sql. Database#to_sql can now take an array of strings and convert into an SQL string. Comments and excessive white-space are removed.

* Improved Schema generator to support data types as method names:
  DB.create_table :test do
    integer :abc
    text :def
    ...
  end

* Implemented ODBC adapter.

=== 0.1.3

* Implemented DBI adapter.

* Refactored database connection code. Now handled through Database#connect.

=== 0.1.2

* The first opened database is automatically assigned to to Model.db.

* Removed SequelConnectionError. Exception class errors are converted to RuntimeError.

* Added support for UNION, INTERSECT and EXCEPT set operations.

* Fixed Dataset#single_record to return nil if no record is found.

* Updated specs to conform to RSpec 1.0.

* Added Model#find_or_create method.

* Fixed MySQL::Dataset#query_single (thanks Dries Harnie.)

* Added Model.subset method. Fixed Model.filter and Model.exclude to accept blocks.

* Added Database#uri method.

* Refactored and removed deprecated code in postgres adapter.

===0.1.1

* More documentation for Dataset.

* Added Dataset#size as alias to Dataset#count.

* Changed Database#<< to call execute (instead of being an alias). Thus it will work for descendants as well.

* Fixed Sequel.open to accept variable arity.

* Refactored Model#refresh, Model.create. Removed Model#reload.

* Refactored Model hooks.

* Cleaned up Dataset API.

=== 0.1.0

* Changed Database#create_table to only accept a block. Nobody's gonna use the other way.

* Removed Dataset#[]= method. Too confusing and not really useful.

* Fixed ConnectionPool#hold to wrap exceptions only once.

* Dataset#where_list Renamed Dataset#expression_list.

* Added support for qualified fields in Proc expressions (e.g. filter {items.id == 1}.)

* Added like? and in? Proc expression operators.

* Added require 'date' in dataset.rb. Is this a 1.8.5 thing?

* Refactored Dataset to use literal strings instead of format strings (slight performance improvement and better readability.)

* Added support for literalizing Date objects.

* Refactored literalization of Time objects.

=== 0.0.20

* Refactored Dataset where clause construction to use expressions.

* Implemented Proc expressions (adapted from a great idea by Sam Smoot.)

* Fixed Model#map.

* Documentation for ConnectionPool.

* Specs for Database.

=== 0.0.19

* More specs for Dataset.

* Fixed Dataset#invert_order to work correctly with strings.

* Fixed Model#== to check equality of values.

* Added Model#exclude and Model#order.

* Fixed Dataset#order and Dataset#group to behave correctly when supplied with qualified field name symbols.

* Removed Database#literal. Shouldn't have been there.

* Added SQLite::Dataset#explain. Returns an array of opcode hashes.

* Specs for ConnectionPool.

=== 0.0.18

* Implemented SequelError and SequelConnectionError classes. ConnectionPool#hold now catches any connection errors and reraises them SequelConnectionError.

* Removed duplication in Database#[].

* :from and :select options are now always arrays (patch by Alex Bradbury.)

* Fixed Dataset#exclude to work correctly (patch and specs by Alex Bradbury.)

=== 0.0.17

* Fixed Postgres::Database#tables to return table names as symbols (caused problem when using Database#table_exists?).

* Fixed Dataset#from to have variable arity, like Dataset#select and Dataset#where (patch by Alex Bradbury.)

* Added support for GROUP BY and HAVING clauses (patches by Alex Bradbury.) Refactored Dataset#filter.

* More specs.

* Refactored Dataset#where for better composability.

* Added Dataset#[]= method.

* Added support for DISTINCT and OFFSET clauses (patches by Alex Bradbury.) Dataset#limit now accepts ranges. Added Dataset#uniq and distinct methods.

=== 0.0.16

* More documentation.

* Added support for subqueries in Dataset#literal.

* Added support for Model.all_by_XXX methods through Model.method_missing.

* Added basic SQL logging to Database.

* Added Enumerable#send_each convenience method.

* Changed Dataset#destroy to return the number of deleted records.

=== 0.0.15

* Improved Dataset#insert_sql to allow arrays as well as hashes.

* Database#drop_table now accepts a list of table names.

* Added Model#id to to return the id column.

=== 0.0.14

* Fixed Model's attribute accessors (hopefully for the last time).

* Changed Model.db and Model.db= to allow different databases for different model classes.

* Fixed bug in aggregate methods (max, min, etc) for datasets using record classes.

=== 0.0.13

* Fixed Model#method_missing to do both find, filter and attribute accessors. duh.

* Fixed bug in Dataset#literal when quoting arrays of strings (thanks Douglas Koszerek.)

=== 0.0.12

* Model#save now correctly performs an INSERT for new objects.

* Added Model#reload for reloading an object from the database.

* Added Dataset#naked method for getting a version of a dataset that fetches records as hashes.

* Implemented attribute accessors for column values ala ActiveRecord models.

* Fixed filtering using nil values (e.g. dataset.filter(:parent_id => nil)).

=== 0.0.11

* Renamed Model.schema to Model.set_schema and Model.get_schema to Model.schema.

* Improved Model class to allow descendants of model clases (thanks Pedro Gutierrez.)

* Removed require 'postgres' in schema.rb (thanks Douglas Koszerek.)

=== 0.0.10

* Added some examples.

* Added Dataset#print method for pretty-printing tables.

=== 0.0.9

* Fixed Postgres::Database#tables and #locks methods.

* Added PGconn#last_insert_id method that should support all 7.x and 8.x versions of Postgresql.

* Added Dataset#exists method for EXISTS where clauses.

* Changed behavior of Dataset#literal to regard symbols as field names.

* Refactored and DRY'd Dataset#literal and overrides therof. Added support for subqueries in where clause.

=== 0.0.8

* Fixed Dataset#reverse_order to provide chainability. This method can be called without arguments to invert the current order or with arguments to provide a descending order.

* Fixed literal representation of literals in SQLite adapter (thanks Christian Neukirchen!) 

* Refactored insert code in Postgres adapter (in preparation for fetching the last insert id for pre-8.1 versions).

=== 0.0.7

* Fixed bug in Model.schema, duh!

=== 0.0.6

* Added Dataset#sql as alias to Dataset#select_sql.

* Dataset#where and Dataset#exclude can now be used for refining dataset conditions, enabling stuff like posts.where(:title => 'abcdef').exclude(:user_id => 3).

* Implemented Dataset#exclude method.

* Added Sequel::Schema#auto_primary_key method for setting an automatic primary key to be added to every table definition. Changed the schema generator to not define a primary key by default.

* Changed Sequel::Database#table_exists? to rely on the tables method if it is available.

* Implemented SQLite::Database#tables.

=== 0.0.5

* Added Dataset#[] method. Refactored Model#find and Model#[].

* Renamed Pool#conn_maker to Pool#connection_proc.

* Added automatic require 'sequel' to all adapters for convenience.

=== 0.0.4

* Added preliminary MySQL support.

* Code cleanup.

=== 0.0.3

* Add Dataset#sum method.

* Added support for exclusive ranges (thanks Christian Neukirchen.)

* Added sequel console for quick'n'dirty access to databases.

* Fixed small bug in Dataset#qualified_field_name for better join support.

=== 0.0.2

* Added Sequel.open as alias to Sequel.connect.

* Refactored Dataset#where_equal_condition into Dataset#where_condition, allowing arrays and ranges, e.g. posts.filter(:stamp => (3.days.ago)..(1.day.ago)), or posts.filter(:category => ['ruby', 'postgres', 'linux']).

* Added Model#[]= method for changing column values and Model#save 
method for saving them.

* Added Dataset#destroy for deleting each record individually as support for models. Renamed Model#delete to Model#destroy (and Model#destroy_all) ala ActiveRecord.

* Refactored Dataset#first and Dataset#last code. These methods can now accept the number of records to fetch.

=== 0.0.1

* More documentation for Dataset.

* Renamed Database#query to Database#dataset.

* Added Dataset#insert_multiple for inserting multiple records.

* Added Dataset#<< as shorthand for inserting records.

* Added Database#<< method for executing arbitrary SQL.

* Imported Sequel code.