Re: Getting into pointers

Hi David,

Although PGM2 may be populating the storage, PGM1 owns (or *should* own) the storage. In other words, PGM1 should be the program that reserves the memory for the structure, and PGM2 should simply populate the data in the structure.

If you do it the other way, then you have to worry about the memory being freed up when PGM2 ends (which starts a whole discussion about activation groups and how they work) and you have to be worried about what happens when more than one program calls PGM2 (do they get the same memory as PGM1? Will PGM1 be affected by the new caller "PGM3" telling PGM2 to change it's data structure, etc)

If PGM1 reserves the storage, and passes it to PGM2, then PGM2 populates it, you'll be doing things correctly. (And mimicking what RPG does for you when you pass a DS as a parameter directly.)


David FOXWELL wrote:

Hi Scott,

My question: Why did you use the BASED keyword? Why bother
using BASED and then %ALLOC? Why not just do this?

D MyDS DS LIKEDS ( gMyOtherDS )
DIM ( 100 )
D MyPtr S *

MyPtr = %addr(MyDS);
PGM2(MyPtr);

Why tell the OS to let you handle the allocation if you're
just going to allocate the exact size of the structure?

I'm afraid the answer to those questions is lack of experience/ignorance.

I feel I'm getting the hang of it though.

I now have something like

D MyDS DS BASED ( MyPtr )
D MyPtr S *

PGM2(MyPtr);

I've got rid of the BASED keyword in PGM2. I've kept it in PGM1 so that MyDS will be usable when PGM2 returns. Is that reasonable?

Can you explain why did you do this :

MyPtr = %addr(MyDS);

Before the call? In my case it's PGM2 that creates the DS and sends back the pointer to PGM1.

Thanks.