On 2/8/06, Simon Belak <[EMAIL PROTECTED]> wrote: > I think passing unexpected values to a method is less of a problem now, > since the method has to be explicitly aware (either via error_handler > decorator, or tg_errors parameter) of the possibility of an invalid value.
True, to a point. There's always a little more complexity right around the corner when you're trying to make something simple. Imagine a widget that lets you pick a person from the database to associate with a record. The web will pass in an integer id, and the method will get an object on the way in. But, if that doesn't validate for some reason, the user will see the form again and that int will still be there. The validator whose job it is to convert the original Person object into an int will get an int instead. That's more of an edge case, though, and can be handled by writing the validator's from_python method usefully and with that understanding. Excepting that, I'm in agreement that it's a lot more pleasant to leave the value untouched. Kevin
