Re: [Haskell-cafe] [Haskell] ANNOUNCE: tardis

On Tue, Aug 7, 2012 at 7:04 AM, Dan Burton wrote:

As a side note, since the code base is relatively small, it can also serve as a simple demonstration of how to use a cabal flag in conjunction with CPP to selectively include swaths of code (see Control/Monad/Tardis.hs and tardis.cabal).

Eep, your API changes based on compile-time settings. I think this is a bad idea, because other packages cannot depend on a flag, so realistically other packages cannot depend on the instances existing, so they're nearly useless. UndecidableInstances is excessively maligned and usually fine anyway. If it compiles, it won't go wrong.