Hi, thanks for the notes.
Yes I agree the current behaviour is more useful, and probably more expected. Note that all of these are true: Rect(0,0,10,10) == [0,0,10,10] [0,0,10,10] == [0,0,10,10] Rect(0,0,10,10) == (0,0,10,10) However: (0,0,10,10) != [0,0,10,10] The rect uses a compare function, where it sees if the other object is a rect like, then compares to it. It definitely does need documenting. I'm not sure if a UserRect style class is useful or not. Cheers, On 6/6/06, Sami Hangaslammi <[EMAIL PROTECTED]> wrote:
On 6/6/06, Rene Dudfield <[EMAIL PROTECTED]> wrote: > Could this behaviour be improved at all? I think the current behaviour is desireable in almost all real use cases. IMHO, Rect should be just a collection of values, so it's logical that rect(0,0,10,10) == rect(0,0,10,10) even if the object instances are different. I think it is enough to document this behaviour, so that those who inherit from Rect don't get any surprises. Another option is to provide another Rect class solely meant for inheritance where you add logic to the object (in which case the object identity usually becomes important). Then it's just a matter of picking a good name for it. RectObject? RectBase? UserRect? I dunno. -- Sami Hangaslammi
