| re: different design considerations for objects and tables
eddiec wrote:
[color=blue]
> hi everyone,
>
> I am designing a pallet management system. Many items are placed onto one
> pallet. The items are all unique and have distinct features. Now, as far
> as application design is concerned the most intuitive way of handling this
> would be to have a Pallet object that holds an array of Item objects,
> which reflects the situation in real life. As far as database design is
> concerned, however, it is easier for the item to 'know' which pallet it
> belongs to.
> e.g. I would create a table for the individual items where one of the
> fields is palletID.
>
> So what do I do? It would be very confusing (I think) to have different
> design approaches in the application and in the database schema.
>
> TIA
>
> eddiec :-)[/color]
Hi,
if you insist on realizing the datastructure within your objects, then this
is a question about OO design, which is of interest in this news group only
as far as C++ happens to support OO design.
However, in C++ there are other approaches to go about your particular
problem. You have two types of objects: pallets and items. Moreover, each
item is placed on one and only one pallet. Thus, abstractly, you have a map
assigning pallets to items:
f : Items --> Paletts.
Viewed from this angle, you might use a container (like
std::map<Item,Palett>) that implements this setup. However, you might also
want to iterate through all items on a given pallet. Thus, the container
should allow for efficient compuatation of preimages and offer an iterator
type for this purpose.
Solving this problem as a (possibly templated) container has the advantage
that you can change the implementation afterwards without affecting the
design of the classes Pallet and Item.
Best
Kai-Uwe Bux
| 
