heap mgr improved?

[rforcen]

yes, that's what i'm currently doing, just using a seq[pointer] to store all 
create calls and finally free then at object destroy, the problem with this 
solution is that it's not elegant and only supports one object at the time