Henning Thielemann wrote:
Is NumLazyByteString a newtype around Bytestring.Lazy that interprets the bit stream represented by the ByteString as integer?
Thomas DuBuisson wrote:
Not exactly. There is not newtype wrapper. NumLazyByteString is: instance Num L.ByteString where ...
Are you absolutely certain that this is the one and only canonical instance that can ever exist that makes sense? Otherwise, please use a newtype wrapper. The reason for this is one of the few remaining major warts in Haskell - that there is absolutely *no way* to prevent the export of instances from a module that you import. So it is considered good practice in current Haskell never to write an instance if anyone has ever published one previously, even if you don't foresee ever using the package in which the instance was defined. Otherwise, someone someday may write code that indirectly depends on both modules in some way, and that code will mysteriously fail to compile. In practice, that implies: never write an unwrapped instance for a datatype that you did not define yourself, and especially not for types that are published and are in common usage, unless you are absolutely certain that your instance is truly canonical and universal. By saying that this is a wart, I am not saying that I think it is a good idea to write more than one instance for the same type intentionally. It is not. But there is no way to prevent it from ever happening inadvertently. So it is important for there to be a mechanism to prevent the import of an unwanted instance when that comes up. Your desire to write this instance is yet another case that demonstrates why this is clearly a wart. Regards, Yitz