2009/7/3 Luke Palmer <lrpal...@gmail.com>: > Keep in mind that with this solution *and* with the > ExistentialQuantification solution, there is no possibility of downcasting. > I.e. if you were planning on making a GraphicalWidget subclass, and them > somewhere seeing if a a Widget is actually a GraphicalWidget, you will be > disappointed. The solution in this case is to redesign your software not to > need downcasting. This is the point at which you are forced to move away > from OO thinking.
Heck, as far as I'm aware most OO communities frown on downcasting too. The "OO approach" is to call a virtual method and let the object decide to do, which (with enough hand-waving) is basically what you end up doing here anyway. Stuart _______________________________________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
