Hi Sergey,


>>I knew the syntax I used was 'sus' but I was surprised at the result. BTW, why bother with the 'This.o=Null' - when wouldn't o go out of scope when c1 is destroyed anyway ?

>I don't want to leave anything to a chance. I consider a good programming practice to excplicitly release such objects.

I know it's generally regarded as good practice (and I tend to conform) but to me it still smacks of fear and superstition <g>. Have you seen a simple concrete (or come to that complicated) example where not nulling an object can cause a problem ?

>>Also I see that although VFP9 (at least in the beta) flags up the error it only does so AFTER creating c2 and is inconsistent in subsequently releasing it.

>Error is only generated if object is assigned to a property this way. Any other type of data doesn't generate an error. I'm not sure what inconsistency you're refering to.

I didn't have much time but I'm sure that in VFP9 whilst the object usually wasn't destroyed until a CLEAR ALL on a couple of attempts it did automatically destroy...

Regards,
Viv