20 Aug 2001 15:46:04 -0600, Alastair David Reid pisze:

An obvious fix would be to change the method name from "typeOf" to "unsafeTypeOf" - reflecting the fact that a bad definition breaks type safety and can lead to segmentation faults. But this wrongly labels uses of typeOf as being unsafe when it's definitions which are (potentially) unsafe.

Wouldn't be possible to make Dynamic derivable or something (better: with a magic default definition of typeOf, so the instance can be made after the type definition)? If the compiler could figure out a good definition itself, it would not only solve the safety problem, but made using Dynamic easier: I always have to look up how a Dynamic instance should look like. Yes, it requires magic in the compiler, but Dynamic is already unimplementable in standard Haskell. -- __("< Marcin Kowalczyk * qrczak@knm.org.pl http://qrczak.ids.net.pl/ \__/ ^^ SYGNATURA ZASTÊPCZA QRCZAK