should -fglasgow-exts support Haskell98?

I don't know if "ghc -fglasgow-exts" should always accept valid
Haskell 98 programs.  If that is the intention, then here is a bug:
  The following program is legal Haskell98 (works fine w/ "ghc") but
  gives a parse error with "-fglasgow-exts":

  let _ #>> _ = 1 in (#>>)

Thanks.

BTW, I've just noticed that the ":t" command in ghci now gives me some
very useful information for values without type signatures (when they
are in a binding group in which other values have signatures): the names
of quantified type variables are nabbed from the other signatures.
This is really nice!

- Mark