To go ahead and be that Haskell purist... in normal Haskell code there are no functions which execute side effects in the process of computing their results. Even a function like
putStrLn :: String -> IO ()
merely produces an IO value. It is only when such a value is sequenced into something presented to the RTS as `main` that the effects are realized.
This seems like pedantry, but it allows you to construct and compose IO values in many ways (besides just sequentially evaluating them). This ends up giving you "macro like" powers.
Thanks for this answer. I think it points clearly to the source of confusion.
To be pedantic back at you, I'm pretty sure your explanation translates to: "So long as you never run a program, Haskell is a pure language."
Pedantically speaking, you can't have it both ways. Either the "Haskell" of the claim "Haskell is pure" includes the RTS or it does not. If it does, it is not pure but can run programs. If it does not, it is pure but cannot run programs.
A programming language that you can't run programs in is not very useful or interesting, so I'd really think it was a better move to define "Haskell" in such a way that programs can be run, and accept that under that definition Haskell is impure, although it retains a kind of effectively pure formalism that allows provability and such, which is awesome.
Either way, it is really important to emphasize in any description of the language that "At runtime Haskell is not pure". Otherwise newbs like me will simply be confused by the contradiction between the claim "Haskell [considered as an unexecutable language] is pure" and "Haskell [considered as a set of instructions to the RTS] is capable of doing IO". The two completely different meanings of "Haskell" must be made clear if profound confusion is to be avoided.
Eh, I don't care much about names. Purity-up-until-RTS being taken seriously makes it easy to think of computations both pure and impure. I get to choose whether to interpret IO as an imperative, impure language subset of Haskell or as a pure type. Both have their advantages.
If your language is impure then you usually can only choose the former.
This seems like pedantry, but it allows you to construct and compose IO values in many ways (besides just sequentially evaluating them). This ends up giving you "macro like" powers.