Re: Pointer usage danger?

On 3/8/11 9:13 AM, M. Lazarus wrote:

Given this code:

D Array1 DS Dim( 10 )
D Field1 2a
D Field2 75a

D Array2 DS Dim( %Elem( Array1 ))
D Based( Array1Ptr ) Inz( %Addr( Array1 ))
D Field1a Like( Field1 )
D Field2a Like( Field2 )

D Array1Ptr s *

/Free
Array1Ptr = %Addr( Array1( 10 )) ;
...

At this point, the first element of Array2 should be positioned over
the last element (#10) of Array1. Assuming I don't attempt to access
element 2 in Array2, is this safe to do or will the shift of Array2,
elements 2-10, overlay memory it shouldn't?

If even the assignment does what is alluded, then that is 'unsafe' because the based storage will indeed "overlay memory it shouldn't". The storage beyond Array1 is undefined, and could occupy effectively random automatic storage. While the source statements may [appear to] avoid addressing elements 2-10, the implementation of those statements may not; e.g. specification on factor ?2? on a call parameter may effect rewrite of the variable, thus writing to storage that was not correctly allocated, which as noted is effectively random storage.

"If even" is noted, because if like I recall in another [PL based] language, the element [literal] specification is ignored when assigning a pointer to an "array". If so in the given scenario, then the effect would have the based Array2 overlay Array1 storage one-for-one in elements, and thus, no longer have the capability to corrupt\address anything other than the declared Array1 data\storage. I did not compile the example to verify the address or see if a warning is issued to that effect; i.e. warn that the element is ignored, and the address of the array rather than an element of the array is used when assigning the address to the pointer.

Regards, Chuck