That seems extremely reasonable. I propose to treat it as
a Haskell98 typo; it doesn't change the meaning of the language
as described by the current report, since no meaning is given
to do {} and friends.
Does anyone think there is a reason *not* to do this?
Simon
> -----Original Message-----
> From: Ross Paterson
> Sent: Monday, July 26, 1999 10:27 AM
> To: [EMAIL PROTECTED]
> Subject: Haskell 98 Report: do expression syntax
>
>
> The Haskell 98 Report has (in 3.14):
>
> exp -> do { stmts } (do expression)
> stmts -> stmt1 ; ... ; stmtn (n>=0)
> stmt -> exp
> | pat <- exp
> | let decls
> | (empty statment)
>
> which allows the following:
>
> do {}
> do {let x = 5}
> do {x <- return 5}
>
> However the translation rules require that the last non-empty stmt
> be an exp, which seems very sensible. I suggest
>
> exp -> do { body } (do expression)
> body -> stmt1 ; ... stmtn ; exp {;} (n>=0)
>
