| Re: Recursive Objects [message #26974] |
Tue, 09 October 2001 13:17  |
David Fanning
Messages: 11724
Registered: August 2001
|
Senior Member |
|
|
David Fanning (david@dfanning.com) writes:
> I've thought all day long that this guy was really
> trying to use an INHERITS in his structure definition,
> not another structure definition. I'm sure this is what
> Karl was thinking, too, in his reply.
Whoops! Of course, I meant "object" definition.
David
--
David W. Fanning, Ph.D.
Fanning Software Consulting
Phone: 970-221-0438, E-mail: david@dfanning.com
Coyote's Guide to IDL Programming: http://www.dfanning.com/
Toll-Free IDL Book Orders: 1-888-461-0155
|
|
|
|
| Re: Recursive Objects [message #26975 is a reply to message #26974] |
Tue, 09 October 2001 13:15  |
David Fanning
Messages: 11724
Registered: August 2001
|
Senior Member |
|
|
Mark Hadfield (m.hadfield@niwa.cri.nz) writes:
> From: "Karl Schultz" <kschultz@researchsystems.com>
>>> pro my_obj__define
>>> struct = {my_obj, another_obj:OBJ_NEW()}
>>> end
>> ....
>> Also note that you do not need to explicitly destroy this instance of
>> "SomeOtherClass" in the my_obj::Cleanup method. Since the objref is in
> the
>> class struct, IDL will find and destroy it for you.
>
> Not in my experience!
>
> When IDL destroys a my_obj instance it will erase an object reference stored
> in the my_obj class structure. It will *not* destroy the heap variable that
> this reference refers to. So to avoid a memory leak you need
>
> pro my_obj::cleanup
> obj_destroy, self.another_obj
> ; Other cleanup tasks
> end
>
I've thought all day long that this guy was really
trying to use an INHERITS in his structure definition,
not another structure definition. I'm sure this is what
Karl was thinking, too, in his reply.
Cheers,
David
--
David W. Fanning, Ph.D.
Fanning Software Consulting
Phone: 970-221-0438, E-mail: david@dfanning.com
Coyote's Guide to IDL Programming: http://www.dfanning.com/
Toll-Free IDL Book Orders: 1-888-461-0155
|
|
|
|
|
|
| Re: Recursive Objects [message #26990 is a reply to message #26976] |
Tue, 09 October 2001 08:51 |
Martin Downing
Messages: 136
Registered: September 1998
|
Senior Member |
|
|
"Francesco Belletti" <guenhwyvar@libero.it> wrote in message
news:e7b4ed30.0110090547.79d8e345@posting.google.com...
> Hello,
> I'm a very beginner with IDL so I'm sorry if my question is too
> stupid.
> I need to store a object reference in an object structure:
>
> pro my_obj__define
> struct = {my_obj, another_obj:OBJ_NEW()}
> end
>
> After another_obj is zeroed during object creation, it couldn't no
> more be used as an object reference!!
> I've tried to redeclare it
>
> another_obj=OBJ_NEW()
>
> but the problem remains.
> It's an IDL bad limit, or my error?
>
> Thank you,
>
> Francesco Belletti
Francesco,
Objects are handled using OBJREFs, which are very much like pointers. What
you have declared is a reference to an object, ie an OBJREF, which you have
set to the NULLOBJ. What you need to do next is to point variable to a valid
object, eg in your INIT method:
----------------------------------------
function my_obj::init, other = other
; see if other is a valid object
if obj_valid(other) then begin
; ok so set pointer to it
self.another_obj = other
print, "setting another"
endif
return, 1
end
function my_obj::GET_ANOTHER
return, self.another_obj
end
pro my_obj__define
struct = {my_obj, another_obj:OBJ_NEW()}
end
-----------------------------------------------
Now when you create some objects:
IDL> obj_a = obj_new('my_obj')
IDL> obj_b = obj_new('my_obj',other =obj_a)
setting another
IDL> help, obj_a
OBJ_A OBJREF = <ObjHeapVar1(MY_OBJ)>
IDL> help, obj_b
OBJ_B OBJREF = <ObjHeapVar2(MY_OBJ)>
IDL> help, obj_b->get_another()
<Expression> OBJREF = <ObjHeapVar1(MY_OBJ)>
The second call to obj_new has not created another copy of obj_a but has
stored its address (OBJREF) within obj_b.
Hope this helps, good luck with IDL!
Martin
|
|
|
|
| Re: Recursive Objects [message #26991 is a reply to message #26990] |
Tue, 09 October 2001 08:49 |
Karl Schultz
Messages: 341
Registered: October 1999
|
Senior Member |
|
|
"Francesco Belletti" <guenhwyvar@libero.it> wrote in message
news:e7b4ed30.0110090547.79d8e345@posting.google.com...
> Hello,
> I'm a very beginner with IDL so I'm sorry if my question is too
> stupid.
> I need to store a object reference in an object structure:
>
> pro my_obj__define
> struct = {my_obj, another_obj:OBJ_NEW()}
> end
This is correct. OBJ_NEW() just puts a null object ref in your struct,
which is what you want at this point.
> After another_obj is zeroed during object creation, it couldn't no
> more be used as an object reference!!
> I've tried to redeclare it
>
> another_obj=OBJ_NEW()
This just puts another null object ref in your structure.
You probably want to create this object in your ::Init method:
function my_obj::Init
self.another_obj = OBJ_NEW('SomeOtherClass')
return, 1
end
Your class "my_obj" is creating an instance of some other object of class
"SomeOtherClass" when it initializes, presumably because your class "my_obj"
needs the services of "SomeOtherClass". Depending on what your "my_obj"
class does, you can instead create the object of class "SomeOtherClass" at
some other time than in the ::Init method, but that all depends on when you
need it.
Also note that you do not need to explicitly destroy this instance of
"SomeOtherClass" in the my_obj::Cleanup method. Since the objref is in the
class struct, IDL will find and destroy it for you.
Karl
|
|
|
|
|
|
| Re: Recursive Objects [message #27085 is a reply to message #26976] |
Wed, 10 October 2001 07:52 |
Karl Schultz
Messages: 341
Registered: October 1999
|
Senior Member |
|
|
"Mark Hadfield" <m.hadfield@niwa.cri.nz> wrote in message
news:001501c150fd$d00b6440$d938a8c0@Hadfield...
> From: "Karl Schultz" <kschultz@researchsystems.com>
>>> pro my_obj__define
>>> struct = {my_obj, another_obj:OBJ_NEW()}
>>> end
>> ....
>> Also note that you do not need to explicitly destroy this instance of
>> "SomeOtherClass" in the my_obj::Cleanup method. Since the objref is in
> the
>> class struct, IDL will find and destroy it for you.
>
> Not in my experience!
>
> When IDL destroys a my_obj instance it will erase an object reference
stored
> in the my_obj class structure. It will *not* destroy the heap variable
that
> this reference refers to. So to avoid a memory leak you need
>
> pro my_obj::cleanup
> obj_destroy, self.another_obj
> ; Other cleanup tasks
> end
Oops, you are right. The object I was thinking about was a container object
(inherits from a container object) where the Init method added an object to
the container right after creating it. So, when the container gets
destroyed, the added object gets destroyed too. Sorry about that.
|
|
|
|
|
|
| Re: Recursive Objects [message #27116 is a reply to message #26974] |
Tue, 09 October 2001 14:03 |
Pavel A. Romashkin
Messages: 531
Registered: November 2000
|
Senior Member |
|
|
David Fanning wrote:
> Whoops! Of course, I meant "object" definition.
Obviously, although David has been silent on the Mac issue, he is a
little shaken, too. Imagine if the only truly supported Windows will be ME?
Cheers,
Pavel
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
