[lnkForumImage]
TotalShareware - Download Free Software

Confronta i prezzi di migliaia di prodotti.
Asp Forum
 Home | Login | Register | Search 


 

Forums >

comp.lang.ruby

Set an instance variable before and after initialize

Martin Jansson

7/22/2006 10:34:00 PM

If possible, I would like to set a instance variable in an object before
and after its initialize method is executed.

I've been looking at Class#new but it doesn't seem to have access to a
reference to the object in creation.

A possible workaround would be to place the object in an already created
hashtable. But I need to know if initialize has been completed or not
even if callcc is used.

This behavior must be shared by all objects in classes that inherit
Object.

I try to create Simula style corutines. In Simula all objects have three
states of execution: active, passive and terminated. An active object is
executing it's class body (think: Rubys Object#initialize on steroids).
The first time an object is activated is when it's created with new.
Within the class body you can call detach to make the object passive and
pause execution of the class body. Execution is then continued after the
point where the object was last activated. Within the class body you can
also call resume wich make the object passive and activates another
object. Execution is continued again when the other object become
passive or terminated. Objects can also become active if the procedure
call is called with the object as an argument. When execution reach the
end of the objects class body, the object becomes terminated and can not
be activated anymore.

You can't do all the things you can do with callcc with this kind of
corutines but it makes some otherwise hairy programming easy.

--
Posted via http://www.ruby-....

11 Answers

Dumaiu

7/23/2006 12:25:00 AM

0

The initialize() method, the constructor, is the earliest point at
which you can manipulate an object, because prior to that it does not
exist. Although I know diddly-doo about coroutines, it sounds to me
like you'd be better off using instances of the hypothetical
"Coroutine" class rather than the class body itself. Let an
initialized Coroutine stay dormant until it is sent a call() message,
or something like that, and then do its yielding and whatever else
Coroutines do.
There seems to be a coroutine implementation, according to Wikipedia,
iff you can read Japanese.

-J

Trans

7/23/2006 1:51:00 AM

0


Martin Jansson wrote:
> If possible, I would like to set a instance variable in an object before
> and after its initialize method is executed.
>
> I've been looking at Class#new but it doesn't seem to have access to a
> reference to the object in creation.

class Klass
def self.new( *args, &blk )
o = allocate
o.instance_variable_set( "@foo", "anything")
o.initialize( *args, &blk )
o.instance_variable_set( "@bar", "anything")
o
end

You could also you #instance_eval { @foo = "anything }

T.

Martin Jansson

7/23/2006 6:57:00 AM

0

Trans wrote:
> Martin Jansson wrote:
>> If possible, I would like to set a instance variable in an object before
>> and after its initialize method is executed.
>>
>> I've been looking at Class#new but it doesn't seem to have access to a
>> reference to the object in creation.
>
> class Klass
> def self.new( *args, &blk )
> o = allocate
> o.instance_variable_set( "@foo", "anything")
> o.initialize( *args, &blk )
> o.instance_variable_set( "@bar", "anything")
> o
> end
>
> You could also you #instance_eval { @foo = "anything }
>
> T.

This break the chain of inheritance. You would have to reimplement all
classes that you use; it would be more complicated to use then Thread
even if it would have the advantage that you could circumvent some of
Threads limitations. I'm not even sure that it would behave correctly.
Class#new seem to do more then call allocate and initialize. But I could
be wrong. Here is the C code for Class#new:

VALUE
rb_class_new_instance(argc, argv, klass)
int argc;
VALUE *argv;
VALUE klass;
{
VALUE obj;

obj = rb_obj_alloc(klass);
rb_obj_call_init(obj, argc, argv);

return obj;
}

I've looked at rb_obj_call_init, but I can't seem to find it at the
moment.

My current hack has an ActiveObject class that I use instead of Object.
In this class I have a method called body instead of initialize.
initialize is already used to set that damn variable and to call body
and can not be rewritten. All this of course breaks the chain of
inherritance.

--
Posted via http://www.ruby-....

Erik Veenstra

7/23/2006 7:36:00 AM

0

That won't work: initialize is private. And it might better to
overwrite Object.new, so it works for all classes.

What about singleton? (I mean "include Singleton".)

gegroet,
Erik V. - http://www.erikve...

----------------------------------------------------------------

class Object
def self.new( *args, &blk )
o = allocate
o.instance_variable_set( "@foo", "anything")
o.instance_eval{initialize( *args, &blk )}
o.instance_variable_set( "@bar", "anything")
o
end
end

class Foo
def initialize(baz)
@baz = baz
end

def bar
p @foo
p @bar
p @baz
end
end

Foo.new(8).bar

----------------------------------------------------------------


Trans

7/23/2006 11:54:00 AM

0

Damn! You guys are sticklers! You want complete, universally
applicable, 100% bug-free implementations for a general mailing list
question. How much time do I have? ;-)

T.

Erik Veenstra

7/23/2006 12:40:00 PM

0

Think about this:

Write once.
Read never.
Use everywhere.

So, yes, "universally applicable" would be nice... ;]

gegroet,
Erik V. - http://www.erikve...


Ara.T.Howard

7/23/2006 2:34:00 PM

0

Ben Nagy

7/26/2006 6:09:00 AM

0

I have a couple of questions...

> -----Original Message-----
> From: ara.t.howard@noaa.gov [mailto:ara.t.howard@noaa.gov]
> Sent: Sunday, July 23, 2006 9:34 PM
> To: ruby-talk ML
> Subject: Re: Set an instance variable before and after initialize
>
> On Sun, 23 Jul 2006, Martin Jansson wrote:
>
> > If possible, I would like to set a instance variable in an
> object before and
> > after its initialize method is executed.
>
> let initialize help you do this then:
>
>
> harp:~ > cat a.rb
> module BeforeAfterInit
> module InstanceMethods
> def before_initialize *a, &b
> end
> def after_initialize *a, &b
> end
> def initialize *a, &b
> before_initialize *a, &b
> r = super

What does this construction (r=super) do? I tried removing both references
to r and replacing it with a single 'super' and it seemed to behave
identically.

> after_initialize *a, &b
> r
> end
> end
> module ClassMethods
> def before &b
> define_method 'before_initialize', &b
> end
> def after &b
> define_method 'after_initialize', &b
> end
> end
> def self.included other
> other.module_eval{ include InstanceMethods }
> other.extend ClassMethods
> super
> end
> end
>
> class C
> include BeforeAfterInit
>
> before{ @a = 40 }
> after{ @b = 2 }

When I define an initialize method here without using super, it breaks
(obvious, I guess), When I use super the before and after code gets run at
the same time - in other words:

def initialize
p @a #should be set, but => nil
super
end

So to my naive understanding this doesn't acheive the goal. What am I
missing?

> def m() @a + @b end
> end
>
> p C.new.m
>
>
>
> harp:~ > ruby a.rb
> 42

Cheers,

ben


Ara.T.Howard

7/26/2006 6:38:00 AM

0

Ben Nagy

7/26/2006 7:01:00 AM

0

> -----Original Message-----
> From: ara.t.howard@noaa.gov [mailto:ara.t.howard@noaa.gov]
[...]
> > def initialize
> > p @a #should be set, but => nil
>
> but why? you haven't called super yet. try calling it first.

Oh, I did that too. My point is that there is no 'before' and 'after' the
code gets run at the same time and it's not possible to insert other code in
between them, which was presumably the point?

def initialize
super
p @b #should not be set yet, initialize method not finished...
end

ben