RE: Dynamic function call

[David Korn <dkorn at pixelpower dot com>]

>-----Original Message-----
>From: Daniel.Andersson@combitechsystems.com
>Sent: 14 June 2001 13:10

> The idéa of my code is that all the object that is created within a thread
>attaches it self to a dispatcher. This is done so that they can receive
>messages that are sent to that thread. Every object attaches it self in a
>list that means that when a message is sent to that position in the list
>then every attached object receives that message.
>
>Doesn't the linker support this kind of dynamic calls?

  Sounds very much like a C++ virtual function call to me.  Nothing to do
with the linker at all.

>-----clip from the attachfunction in dispatcher-----
>
>void CDispatcher::Attach(CCommon* pMsgHandler, Receiver_t RecObj)
>{
> AttachedObjects[RecObj] = pMsgHandler;
>}
>----endclip------

  Ok, so for each receiver type, you have an array entry that points to
the object that wants to receive the messages.  I got the impression from
the way you described this above that you wanted a *list* of objects of a
particular type, so they could all receive the message, but you could easily
extend that code to make a singly-linked list, using a next item pointer in
the base class CCommon.

>-----clip from the caller of dispatcher-----
>Dispatcher.Attach(this, USERS);
>----endclip------

  See, that will be the only object that gets messages for USERS: if there
was already one registered, it will be overwritten in the array. 

>Here comes all the different ways that i have tried so far. I want to be
>able to use the first one but is doesn't work. The array AttachedObjects
>declared as a CBase* AttachedObjects[N] so it will contain of a lot of

  CBase * ? Do you not mean CCommon * ?  Or is CCommon derived from CBase,
and the other class types (CUsers, CCar, CNodeman ?) derived from CCommon ?

>pointers to all the attached objects. None of the three alternatives below
>works!! The only thing that works is when i hardcode the object name in the
>function (see clip below)
>
>Any hints?

  It would have been good if you'd told us in what way it doesn't work.
Does the code fail to compile, or does it just do the wrong thing when
it executes ?

>Alt 1
>AttachedObjects[(msgRaw->Receiver)]->EventHandler(msgRaw);

  Looking at your code, I would expect that it does compile (although you
should have shown us the definition of EventHandler), but runs wrong.  And
here's the problem: all the pointers in the AttachedObjects array are of
type CCommon *, so the function that gets called here is
CCommon::EventHandler.  What you really want I assume is for it to call
(type derived from CCommon)::EventHandler, but the compiler can't do that
by default, because it doesn't know what type of object the array entry
points to: it knows that it points to a CCommon object, but it doesn't know
that that might be the CCommon base part of a derived object.

  This is exactly the sort of situation that virtual functions were made
for!  If you declare EventHandler to be a virtual function in the CCommon
class definition, that means that every object of type CCommon, and every
object of a type derived from CCommon, will carry around a pointer to the
EventHandler function for that particular type.  This means that when the
only pointer you have to the object is of CCommon * type, the compiler will
still be able to compile code to get to the correct function, by looking
up the function's address from the actual object itself.

  For more info, refer to a C++ textbook on the subject of virtual
functions.

          DaveK
-- 
we are not seats or eyeballs or end users or consumers.
we are human beings - and our reach exceeds your grasp.
                    deal with it.                      - cluetrain.org 


**********************************************************************
This email and any files transmitted with it are confidential and
intended solely for the use of the individual or entity to whom they
are addressed. If you have received this email in error please notify
the system manager.

This footnote also confirms that this email message has been swept by
MIMEsweeper for the presence of computer viruses.

www.mimesweeper.com
**********************************************************************

------
Want more information?  See the CrossGCC FAQ, http://www.objsw.com/CrossGCC/
Want to unsubscribe? Send a note to crossgcc-unsubscribe@sourceware.cygnus.com