On 2006-08-02, Diez B. Roggisch <[EMAIL PROTECTED]> wrote: >> I find that a strange purpose because when you are working on a class, >> you don't necessarily know if you will ever know many instance of that >> class. So should I use __slots__ in all my classes, just to be sure >> for when someone wants many instances of one? > > I find that a strange reasoning because when you are working on a class, > you don't necessarily know if you will ever know if it needs a > __getitem__-method. > So do you use __getitem__ in all your classes, just to be sure > for when someone wants __getitem__ in one?
That is IMO a totally different issue. The __getitem__ decision is an API decision. That is normally somthing the implementor decides. However it is the user of a class that decides how many instances he needs of that class. > To my experience programming often means that requirements change - and one > has to adapt. Who has to adapt? The user or the writer of the class? If I'm using a class written by someone, I sure wouldn't like to have to go through it because of memory issues. > If memory becomes an issue, you might overcome it using > slots. Or a disk cache. Or buy new memory. Does that imply for you that you > buy new disks and memory each time you start coding? I would think that someone writing a class, shouldn't have his instances use more memory than they need. -- Antoon Pardon -- http://mail.python.org/mailman/listinfo/python-list
