Disposing generic pointers

John L. Ries jries at salford-systems.com
Tue Apr 6 18:47:34 CEST 2010

On Tue, 6 Apr 2010, Adriaan van Os wrote:

> John L. Ries wrote:
>> On Tue, 6 Apr 2010, Adriaan van Os wrote:
>>> John L. Ries wrote:
>>>> Since the unit in question is supposed to work with whatever data type
>>>> is thrown at it (it is intended to be generic set of queuing routines),
>>> But how are the queue elements allocated ?
>> By the calling program.  All the queue unit ever sees are pointers and
>> its own infrastructure.
> But then, I think it would be good style to add a call to the unit that 
> does the allocation - if you really want the unit to be generic, as you 
> write. After all, it is possible that the calling program chooses an 
> allocater that is incompatible with the deallocator in the unit.
Perhaps.  In general, it is the calling program that is expected to 
deallocate the elements (after dequeueing them).  The procedure I sent in 
my initial message is intended to delallocate a queue that might not be 
empty, though normally it would be (but it is still a good idea to check). 
It is actually the only routine in the unit that deallocates elements. 
Possibly, for Borland compatibility, I will need to add an argument to the 
procedure that gives the element length, but that is a lot more C-ish than 
I like.

John L. Ries              |
Salford Systems           |
Phone: (619)543-8880 x107 |
or     (435)867-8885      |

More information about the Gpc mailing list