-
Fix ActiveRecord 2.0.0 - 2.2.3 integrations failing if version info isn’t already loaded
-
Fix integration with dirty attribute tracking on DataMapper 0.10.3
-
Fix observers failing in ActiveRecord 3.0.0.beta4+ integrations
-
Fix deprecation warning in Rails 3 railtie [Chris Yuan]
-
Use attribute-based event transitions whenever possible to ensure consistency
-
Fix action helpers being defined when the action is only defined in the machine’s owner class
-
Disable attribute-based event transitions in DataMapper 0.9.4 - 0.9.6 when dm-validations is being used
-
Add support for DataMapper 0.10.3+
-
Add around_transition callbacks
-
Fix transition failures during save not being handled correctly in Sequel 2.12.0+
-
Fix attribute-based event transitions not hooking in properly in DataMapper 0.10.0+ and Sequel 2.12.0+
-
Fix dynamic initial states causing errors in Ruby 1.9+ if no arguments are defined in the block
-
Add MongoMapper 0.5.5+ support
-
Add ActiveModel 3.0+ support for use with integrations that implement its interface
-
Fix DataMapper integration failing when ActiveSupport is loaded in place of Extlib
-
Add version dependencies for ruby-graphviz
-
Remove app-specific rails / merb rake tasks in favor of always running state_machine:draw
-
Add Rails 3 railtie for automatically loading rake tasks when installed as a gem
-
Release gems via rake-gemcutter instead of rubyforge
-
Move rake tasks to lib/tasks
-
Dispatch state behavior to the superclass if it’s undefined for a particular state [Sandro Turriate and Tim Pope]
-
Fix state / event names not supporting i18n in ActiveRecord
-
Fix original ActiveRecord::Observer#update not being used for non-state_machine callbacks [Jeremy Wells]
-
Add support for ActiveRecord 3.0
-
Fix without_{name} scopes not quoting columns in ActiveRecord [Jon Evans]
-
Fix without_{name} scopes not scoping columns to the table in ActiveRecord and Sequel [Jon Evans]
-
Fix custom state attributes not being marked properly as changed in ActiveRecord
-
Fix tracked attributes changes in ActiveRecord / DataMapper integrations not working correctly for non-loopbacks [Joe Lind]
-
Fix plural scope names being incorrect for DataMapper 0.9.4 - 0.9.6
-
Fix deprecation warnings for ruby-graphviz 0.9.0+
-
Add support for ActiveRecord 2.0.*
-
Fix nil states being overwritten when they’re explicitly set in ORM integrations
-
Fix default states not getting set in ORM integrations if the column has a default
-
Fix event transitions being kept around while running actions/callbacks, sometimes preventing object marshalling
-
Add support for DataMapper 0.10.0
-
Always interpet nil return values from actions as failed attempts
-
Fix loopbacks not causing records to save in ORM integrations if no other fields were changed
-
Fix events not failing with useful errors when an object’s state is invalid
-
Use more friendly NoMethodError messages for state-driven behaviors
-
Fix before_transition callbacks getting run twice when using event attributes in ORM integrations
-
Add the ability to query for the availability of specific transitions on an object
-
Allow after_transition callbacks to be explicitly run on failed attempts
-
By default, don’t run after_transition callbacks on failed attempts
-
Fix not allowing multiple methods to be specified as arguments in callbacks
-
Fix initial states being set when loading records from the database in Sequel integration
-
Allow static initial states to be set earlier in the initialization of an object
-
Use friendly validation errors for nil states
-
Fix states not being validated properly when using custom names in ActiveRecord / DataMapper integrations
-
Allow multiple state machines on the same class to target the same attribute
-
Add support for :attribute to customize the attribute target, assuming the name is the first argument of #state_machine
-
Simplify reading from / writing to machine-related attributes on objects
-
Fix locale for ActiveRecord getting added to the i18n load path multiple times [Reiner Dieterich]
-
Fix callbacks, guards, and state-driven behaviors not always working on tainted classes [Brandon Dimcheff]
-
Use Ruby 1.9’s built-in Object#instance_exec for bound callbacks when it’s available
-
Improve performance of cached dynamic state lookups by 25%
-
Add built-in caching for dynamic state values when the value only needs to be generated once
-
Fix flawed example for using record ids as state values
-
Don’t evaluate state values until they’re actually used in an object instance
-
Make it easier to use event attributes for actions defined in the same class as the state machine
-
Fix #save/save! running transitions in ActiveRecord integrations even when a machine’s action is not :save
-
Fix #save! not firing event attributes properly in ActiveRecord integrations
-
Fix log files being included in gems
-
Require DataMapper version be >= 0.9.4
-
Explicitly load Sequel’s built-in inflector (>= 2.12.0) for scope names
-
Don’t use qualified name for event attributes
-
Fix #valid? being defined for DataMapper resources when dm-validations isn’t loaded
-
Add auto-validation of values allowed for the state attribute in ORM integrations
-
Add support for running multiple methods in a callback without using blocks
-
Add more flexibility around how callbacks are defined
-
Add security documentation around mass-assignment in ORM integrations
-
Fix event attribute transitions being publicly accessible
-
Fix machines failing to generate graphs when run from Merb tasks
-
Add #{attribute}_event for automatically firing events when the object’s action is called
-
Make it easier to override state-driven behaviors
-
Rollback state changes when the action fails during transitions
-
Use :messages instead of :invalid_message for customizing validation errors
-
Use more human-readable validation errors
-
Add support for more ActiveRecord observer hooks
-
Add support for targeting multiple specific state machines in DataMapper observer hooks
-
Don’t pass the result of the action as an argument to callbacks (access via Transition#result)
-
Fix incorrect results being used when running transitions in parallel
-
Fix transition args not being set when run in parallel
-
Allow callback terminators to be set on an application-wide basis
-
Only catch :halt during before / after transition callbacks
-
Fix ActiveRecord predicates being overwritten if they’re already defined in the class
-
Allow machine options to be set on an integration-wide basis
-
Turn transactions off by default in DataMapper integrations
-
Add support for configuring the use of transactions
-
Simplify reading/writing of attributes
-
Simplify access to state machines via #state_machine(:attribute) without generating dupes
-
Fix assumptions that dm-validations is always available in DataMapper integration
-
Automatically define DataMapper properties for machine attributes if they don’t exist
-
Add Transition#qualified_event, #qualified_from_name, and #qualified_to_name
-
Add #fire_events / #fire_events! for running events on multiple state machines in parallel
-
Rename next_#{event}_transition to #{event}_transition
-
Add #{attribute}_transitions for getting the list of transitions that can be run on an object
-
Add #{attribute}_events for getting the list of events that can be fired on an object
-
Use generated non-bang event when running bang version so that overriding one affects the other
-
Provide access to arguments passed into an event from transition callbacks via Transition#args
-
Add support for customizing the graph’s orientation
-
Use the standard visualizations for initial (open arrow) and final (double circle) states
-
Highlight final states in GraphViz drawings
-
Make it easier to override generated instance / class methods
-
Add i18n support for ActiveRecord validation errors
-
Add a validation error when failing to transition for ActiveRecord / DataMapper / Sequel integrations
-
Allow multiple conditions for callbacks / class behaviors
-
Add support for state-driven class behavior with :if/:unless options
-
Alias Machine#event as Machine#on
-
Fix nil from/to states not being handled properly
-
Simplify hooking callbacks into loopbacks
-
Add simplified transition/callback requirement syntax
-
Improve pretty-print of events
-
Simplify state/event matching design, improving guard performance by 30%
-
Add better error notification when conflicting guard options are defined
-
Fix scope name pluralization not being applied correctly
-
Allow states to be drawn as ellipses to accommodate long names
-
Fix rake tasks not being registered in Rails/Merb applications
-
Never automatically define machine attribute accessors when using an integration
-
Add to_name and from_name to transition objects
-
Add nicely formatted #inspect for transitions
-
Fix ActiveRecord integrations failing when the database doesn’t exist yet
-
Fix states not being drawn in GraphViz graphs in the correct order
-
Add nicely formatted #inspect for states and events
-
Simplify machine context-switching
-
Store events/states in enumerable node collections
-
No longer allow subclasses to change the integration
-
Move fire! action logic into the Event class (no longer calls fire action on the object)
-
Allow states in subclasses to have different values
-
Recommend that all states be referenced as symbols instead of strings
-
All states must now be named (and can be associated with other value types)
-
Add support for customizing the actual stored value for a state
-
Add compatibility with Ruby 1.9+
-
Allow dm-observer integration to be optional
-
Fix non-lambda callbacks not working for DataMapper/Sequel
-
Fix graphs not being drawn the same way consistently
-
Add support for sharing transitions across multiple events
-
Add support for state-driven behavior
-
Simplify initialize hooks, requiring super to be called instead
-
Add :namespace option for generated state predicates / event methods
-
Fix nil states not being handled properly in guards, known states, or visualizations
-
Fix the same node being used for different dynamic states in GraphViz output
-
Always include initial state in the list of known states even if it’s dynamic
-
Use consistent naming scheme for dynamic states in GraphViz output
-
Allow blocks to be directly passed into machine class
-
Fix attribute predicates not working on attributes that represent columns in ActiveRecord
-
Remove the PluginAWeek namespace
-
Add generic attribute predicate (e.g. “#{attribute}?(state_name)”) and state predicates (e.g. “#{state}?”)
-
Add Sequel support
-
Fix aliasing :initialize on ActiveRecord models causing warnings when the environment is reloaded
-
Fix ActiveRecord state machines trying to query the database on unmigrated models
-
Fix initial states not getting set when the current value is an empty string [Aaron Gibralter]
-
Add rake tasks for generating graphviz files for state machines [Nate Murray]
-
Fix initial state not being included in list of known states
-
Add other_states directive for defining additional states not referenced in transitions or callbacks [Pete Forde]
-
Add next_#{event}_transition for getting the next transition that would be performed if the event were invoked
-
Add the ability to override the pluralized name of an attribute for creating scopes
-
Add the ability to halt callback chains by: throw :halt
-
Add support for dynamic to states in transitions (e.g. :to => lambda {Time.now})
-
Add support for using real blocks in before_transition/after_transition calls instead of using the :do option
-
Add DataMapper support
-
Include states referenced in transition callbacks in the list of a machine’s known states
-
Only generate the known states for a machine on demand, rather than calculating beforehand
-
Add the ability to skip state change actions during a transition (e.g. vehicle.ignite(false))
-
Add the ability for the state change action (e.g.
savefor ActiveRecord) to be configurable -
Allow state machines to be defined on any Ruby class, not just ActiveRecord (removes all external dependencies)
-
Refactor transitions, guards, and callbacks for better organization/design
-
Use a class containing the transition context in callbacks, rather than an ordered list of each individual attribute
-
Add without_#{attribute} named scopes (opposite of the existing with_#{attribute} named scopes) [Sean O’Brien]
-
Fix the initial state not getting set when the state attribute is mass-assigned but protected
-
Change how the base module is included to prevent namespacing conflicts
-
No longer allow additional arguments to be passed into event actions
-
Add support for can_#{event}? for checking whether an event can be fired based on the current state of the record
-
Don’t use callbacks for performing transitions
-
Fix state machines in subclasses not knowing what states/events/transitions were defined by superclasses
-
Replace all before/after_exit/enter/loopback callback hooks and :before/:after options for events with before_transition/after_transition callbacks, e.g.
before_transition :from => ‘parked’, :do => :lock_doors # was before_exit :parked, :lock_doors after_transition :on => ‘ignite’, :do => :turn_on_radio # was event :ignite, :after => :turn_on_radio do
-
Always save when an event is fired even if it results in a loopback [Jürgen Strobel]
-
Ensure initial state callbacks are invoked in the proper order when an event is fired on a new record
-
Add before_loopback and after_loopback hooks [Jürgen Strobel]
-
Add more descriptive exceptions
-
Assume the default state attribute is “state” if one is not provided
-
Add :except_from option for transitions if you want to blacklist states
-
Add PluginAWeek::StateMachine::Machine#states
-
Add PluginAWeek::StateMachine::Event#transitions
-
Allow creating transitions with no from state (effectively allowing the transition for any from state)
-
Reduce the number of objects created for each transition
-
Add a non-bang version of events (e.g. park) that will return a boolean value for success
-
Raise an exception if the bang version of events are used (e.g. park!) and no transition is successful
-
Change callbacks to act a little more like ActiveRecord
-
Avoid using string evaluation for dynamic methods
-
Remove log files from gems
-
Completely rewritten from scratch
-
Renamed to state_machine
-
Removed database dependencies
-
Removed models in favor of an attribute-agnostic design
-
Use ActiveSupport::Callbacks instead of eval_call
-
Remove dry_transaction_rollbacks dependencies
-
Added functional tests
-
Updated documentation
-
Add dependency on custom_callbacks
-
Move test fixtures out of the test application root directory
-
Improve documentation
-
Remove the StateExtension module in favor of adding singleton methods to the stateful class
-
Convert dos newlines to unix newlines
-
Fix error message when a given event can’t be found in the database
-
Add before_#{action} and #{action} callbacks when an event is performed
-
All state and event callbacks can now explicitly return false in order to cancel the action
-
Refactor ActiveState callback creation
-
Refactor unit tests so that they use mock classes instead of themselves
-
Allow force_reload option to be set in the state association
-
Don’t save the entire model when updating the state_id
-
Raise exception if a class tries to define a state more than once
-
Add tests for PluginAWeek::Has::States::ActiveState
-
Refactor active state/active event creation
-
Fix owner_type not being set correctly in active states/events of subclasses
-
Allow subclasses to override the initial state
-
Fix problem with migrations using default null when column cannot be null
-
Moved deadline support into a separate plugin (has_state_deadlines).
-
Added many more unit tests.
-
Simplified many of the interfaces for maintainability.
-
Added support for turning off recording state changes.
-
Removed the short_description and long_description columns, in favor of an optional human_name column.
-
Fixed not overriding the correct equality methods in the StateTransition class.
-
Added to_sym to State and Event.
-
State#name and Event#name now return the string version of the name instead of the symbol version.
-
Added State#human_name and Event#human_name to automatically figure out what the human name is if it isn’t specified in the table.
-
Updated manual rollbacks to use the new Rails edge api (ActiveRecord::Rollback exception).
-
Moved StateExtension class into a separate file in order to help keep the has_state files clean.
-
Renamed InvalidState and InvalidEvent exceptions to StateNotFound and EventNotFound in order to follow the ActiveRecord convention (i.e. RecordNotFound).
-
Added StateNotActive and EventNotActive exceptions to help differentiate between states which don’t exist and states which weren’t defined in the class.
-
Added support for defining callbacks like so:
def before_exit_parked end
def after_enter_idling end
-
Added support for defining callbacks using class methods:
before_exit_parked :fasten_seatbelt
-
Added event callbacks after the transition has occurred (e.g. after_park)
-
State callbacks no longer receive any of the arguments that were provided in the event action
-
Updated license to include our names.