RE: [ECOS] A C++ question regarding the delete operator

["Chris Sekula" <chriss at turnpikeglobal dot com>]

Thanks for your response. Your interpretation of my question (an
automatic variable on the stack) was correct. 
 
CHRIS.

	-----Original Message----- 
	From: Rosimildo da Silva 
	Sent: Tue 1/1/2002 11:57 AM 
	To: Chris Sekula 
	Cc: ecos-discuss@sources.redhat.com 
	Subject: Re: [ECOS] A C++ question regarding the delete operator
	
	

	From: "Chris Sekula" <chriss@turnpikeglobal.com>
	To: "Rosimildo da Silva" <rosimildo@hotmail.com>
	Cc: <ecos-discuss@sources.redhat.com>
	Sent: Tuesday, January 01, 2002 10:02 AM
	Subject: RE: [ECOS] A C++ question regarding the delete operator
	
	
	> Yes, after looking through the code a bit more thoroughly, I
believe
	> that memory had been allocated in another method when the
pointer was
	> created.
	
	OK.
	
	> However, I'm still curious as to what may happen if 'delete'
is called
	> on a pointer that is created locally without the 'new'
operator ever
	> being invoked.
	
	I do not understand what you mean by "created locally", if you
mean
	an automatic variable on stack, such as:
	
	void foo()
	{
	    MyClass  x;
	
	    // some code
	
	    MyClass *px = &x;  
	
	    // more stuff
	
	    // THIS PROBABLY CAUSES A CRASH, OR EVEN DESTROY
	    // THE HEAP. WORST, THE PROGRAM MOST LIKELY WILL
	    // NOT CRASH NOW, BUT LATER WHEN ANOTHER POINT
	    // TRY TO ACCESS THE HEAP OR MEMORY POOLS, ETC.
	    delete px;
	
	}
	
	Botton line: Only use "delete" for things allocated with "new"
	
	   new ---------> free     ==>  BAD, VERY BAD
	   malloc-------> delete  ==>  BAD, VERY BAD
	
	
	Do not read below:
	-------------------
	
	There is a way to "initialize" an object at a given "buffer". It
is called
	"placement new", but you do not want deal with this unless you
have
	at least 5 years of C++ ( doing it everyday !!! <g> ).
	
	Rosimildo.