rubyist / aasm

I have a class:

class Item < ActiveRecord::Base
  include AASM
  aasm_initial_state :available
  aasm_state :available
  aasm_state :in_use
  aasm_event(:check_out) do
    transitions :from => :available, :to => :in_use
  end
  aasm_event(:check_in) do
    transitions :from => :in_use, :to => :available
  end
end

This works:

i = Item.create!
i.write_attribute(:state, 'in_use')
# => "in_use"
i.save!
# => true
i.state
# => "in_use"

But this doesn't do anything:

i = Item.create!
i.check_out!
# => true
i.state
# => "available"

Hi, the problem i've described in subj. Possible - it may be solution - http://github.com/railsbob/aasm/commit/faa145834614bd9d4a02f8c114fc736d26e1323e

I need an 'after' call back just like acts_as_state_machine. It is posable that a callback may need to transition to a new state, currently that isn't posable since the state transition hasn't completed yet when the 'enter' callback is called. This might happen for example when you call a pay! action, and may end up with a 'cleared' or 'decline' state.

aasm_column :state aasm_initial_state :pending
aasm_state :pending aasm_state :current, :enter => :deliver aasm_state :old, :enter => :end_now

aasm_event :activate do

transitions :from => :pending, :to => :current

end

aasm_event :expire do

transitions :from => :current, :to => :old

end

-------------- RSpec:

it "should transition from pending to current with activate! event" do

@my_obj.activate!
@my_obj.state.should eql('current')

end

it "should transition from current to old with expire! event" do

@my_obj.activate!
@my_obj.state.should eql('current')
@my_obj.expire!
@my_obj.state.should eql('old')

end

Although the expire even calls the callback 'end_now', it never transitions the model to the 'old' state. While on the contrary activate! does transition the state correctly. Removing the callback from 'old' doesn't help. I can't see any reason this would fail.

Hello,

I don't actually know that much about AASM, but it's part of a project I'm involved in that I wanted to graph using railroad. Unfortunately, railroad crashes when it processes AASM - but it appears to be due to a bug in AASM. There seem to be lots of people wanting to do something similar. I believe I've identified the problem, and a solution that at least fixes the railroad crash, so I thought I would let you know.

When railroad runs against projects that use AASM, it crashes with a stack overflow. I don't really understand the AASM code, so I can't really tell if this could happen under normal operations. I've identified the "bug" and a "fix." The fix seems safe, but it's definitely a band-aid. If the problem can't come up in normal operation, then the band-aid is probably sufficient.

The problem occurs on line 242 of active_record_persistence.rb (as for versions, I don't use git, but FWIW, the file I downloaded is called rubyist-aasm-7a1c70158c67113bc1c8dd6bd6dbd327af150648):

 module NamedScopeMethods
   def aasm_state_with_named_scope name, options = {}

  aasm_state_without_named_scope name, options
     self.named_scope name, :conditions => { "#{table_name}.#{self.aasm_column}" => name.to_s} unless self.respond_to?(name)
   end
 end

The problem seems to be that for some reason, when railroad tries to process AASM code, the aliasing earlier in this file causes the above method to call itself:

(lines 44-50)

     base.class_eval do
       class << self
         unless method_defined?(:aasm_state_without_named_scope)
           alias_method :aasm_state_without_named_scope, :aasm_state
           alias_method :aasm_state, :aasm_state_with_named_scope
         end
       end

The band-aid fix is to put a limiter on the call like so:

     aasm_state_without_named_scope name, options.merge(:called_as_unnamed => true) unless options[:called_as_unnamed]

I believe this is safe because it seems to me that in any case, if this line is calling the function that contains it, then it's going to crash anyway.

-Avram

Hi,

I have an app that uses a form-wizard that allows you to go forwards/backawards to form pages. The logic to keep up with the current page is a bit clunky and it sounds like aasm might be a good improvement. I would use a form page as a 'state', then use aasm to track what the next/previous pages are.

However, admin users can create form pages dynamically and these get stored in the database via ActiveRecord. How would you deal with this in aasm, where states are typically static?

Kim

Is it intentional that the :enter method is not called for the initial state?
(i.e. if this is intentional, let me know and I'll update the README)

e.g. with the following AASM definition, the do_create is not called on the initial creation of the object.

include AASM
aasm_column :state
aasm_initial_state :pending

aasm_state :pending, :enter => :do_create
aasm_state :active, :enter => :do_activate

aasm_event :activate do
  transitions :to => :active, :from => [:pending]
end

aasm_event :deactivate do
  transitions :to => :pending, :from => [ :active ]
end

When using the :on_transition option to transitions argument passing requires an extra parameter

aasm_event :something do

transitions :from=>:state1, :to=>:state2, :on_transition=>:foo

end

def foo(a,b)
...... end

In this case for foo to be called with a=1 and b=2 when the transition occurs we need to call
something(nil,1,2) or something(:state2,1,2) - This is because the first parameter is taken as the destination state. This seems superfluous when there is only one destination state - Could this be changed to support a syntax like something(1,2) in the case of only one destination state or something(1,2,:to=>:state2) in the case of multiple possible destinations?

Please, change in README.rdoc
oldstate:after_enter
to
newstate:after_enter