Plan A does seem better.

Speaking of making changes to Typeable, is there a chance of connecting/converting between Data.Typeable.TypeRep and Language.Haskell.TH.Type?

Oren Ben-Kiki

On Wed, Oct 3, 2012 at 3:03 PM, Edward Kmett <ekmett@gmail.com> wrote:
I'm definitely in favor of Plan A.

-Edward

On Wed, Oct 3, 2012 at 8:01 AM, Simon Peyton-Jones <simonpj@microsoft.com> wrote:
* If we do make Typeable instances by default, there are two paths
   Plan A (easy): make GHC derive Typeable for everything, deprecate
            all uses of 'deriving Typeable'.  Small downside: some programs
            that are currently rejected will be accepted.

   Plan B (tiresome): have AutoDeriveTypable as an extension.  That
           means maintaining the massive instance table.

   Personally I vote for Plan A.   It's easier and more efficient to implement,
   and the upside of Plan B is modest in the extreme.