I'd like to complain bitterly...

from Thorsten's Epigram course, (trying not to give the game away :) ):

----------------------------------------------------------------------------------------------------------------------------------
     ( P : A -> * !        ( P : A -> * ;  a : A ;  p : P a !
data !------------!  where !--------------------------------!
     !  Ex P : *  )        !        wit P a p : Ex P        )
----------------------------------------------------------------------------------------------------------------------------------
...
----------------------------------------------------------------------------------------------------------------------------------
     ( P, Q : A -> * ;  p : Ex (lam a => Or (P a) (Q a)) !
let  !---------------------------------------------------!
     !             ex8 p : Or (Ex P) (Ex Q)              )
                                                          
     ex8 p <= case p                                      
     { ex8 (wit (lam a' => Or ( a') (' a')) a p) []       
     }                                                    
----------------------------------------------------------------------------------------------------------------------------------

That is the 'helpful' name choices (for the things that are called P
and Q in the type) are *nothing* and prime, personally i prefered x4
and x5 the i got from the older version. They certanly make it easier
to refer to the first of the two...

Peter


On Mon, 04 Oct 2004 17:40:27 +0100, Conor McBride <[EMAIL PROTECTED]> wrote:
> Hi folks
> 
> I realise it's hardly the most vital change, but I thought I'd try to
> get back under the bonnet with a relatively easy job.
> 
> I've just committed a version which is a wee bit better at choosing
> names when you use the <= rule. In particular, rec doesn't automatically
> rename its argument to x.
> 
> The idea is basically this: beta-0 does alpha. What? When you instantiate
> a bound variable x with another variable y, that induces a renaming on the
> other bound variables in x's scope, based on matching x's name with y's
> name. Naturally, this doesn't change the meaning of that scope; it just
> makes the default names a bit more suggestive.
> 
> Try it and see...whenever it propagates, which will be sooner or later, I
> suppose. Come back and complain bitterly if it makes things worse.
> 
> Cheers
> 
> Conor
> 
> --
> http://www.cs.rhul.ac.uk/~conor
>

Reply via email to