71b
Subj : Re: Can you mimic atexit() for class functions?
To   : borland.public.cpp.borlandcpp
From : Jeff Kish <jeff.kish@mro.com>
Date : Wed Oct 08 2003 09:56 am

On Tue, 7 Oct 2003 23:08:32 -0400, "Ed Mulroy [TeamB]"
<dont_email_me@bitbuc.ket> wrote:

>Classes are gone by the time atexit does its thing.
>
>An AddCallback function shouldn't be too difficult to write, but what you
>describe sounds as if it is supposed to be called at the end of the program.
>The way to do that with classes is to place either the code for what you
>want or the code for the call in the class destructor.  If it needs to run
>at the end of the program, then write a class and declare a global instance
>of it and put the code in the destructor.
>
>.  Ed
>
>> gary wrote in message
>> news:3F837C67.3040604@cordelli.net...
>>
>> just wondering if it is possible to have a member function in various
>> (different) classes, but provide a single additional class that defines
>> some kind of "AddCallback()" function that each object of the various
>> other classes can call in order to add their own member function to a
>> callback list so that the single additional class can invoke the list of
>> added callbacks each in turn ?
>>
>> this is sort of like the way atexit() works in C, but in that case it is
>> easy to just define every added callback as "void (*pfn)()", whereas in
<SNIP>
Hey, don't forget, I don't thinkg there is any guarantee as far as the
order of creation/destruction goes for static/global objects.
I'm right, aren't I? Also the order of dlls and how they get shut down
is unpredictable also, isn't it? I think this makes final cleanup
painful sometimes.
I appreciate any clarification/tutorial/"you ought to know better"
type information here.

Thanks
Jeff Kish

.

0