[Haskell] GADTs and GHC

Simon Peyton-Jones simonpj at microsoft.com
Fri Oct 14 11:35:24 EDT 2005

Dear Haskell folk

One of the interesting things about ICFP was the number of times GADTs
came up, either in papers or in discussions in the breaks.   Francois's
invited talk was very inspiring, and that was just the beginning.

Stephanie and Dimitrios and I are now busy revising our wobbly-types
GADT paper (which is implemented in GHC, but never published
http://research.microsoft.com/%7Esimonpj/papers/gadt/index.htm).  We
have a *far* simpler story now... but that means it'll reject some
programs that would be accepted by the earlier version.  We don't think
this loss of expressiveness is important, but we'd like to check.  Hence
this message:

	If you use GADTs, would you to send us your GADT code, 
	so we can check that we can still compile it?  

(A secondary reason for asking is that it'd be interesting just to see
the way you are using GADTs.)  

If you send your code just to me, I'll collect it.  You could

a) send a complete Haskell program, which should compile and run

b) send a Haskell module, or collection of modules, which should
	but which isn't a complete program

c) send a Haskell module using GADTs,  but lacking the support modules
	that would enable it to compile

d) send Omega code, which GHC won't understand, but which
	demonstrates some cunning use of GADTs

All of these would be useful, even (c) and (d) --- we can eyeball the
code.  As a bonus, in the case of (a) or (b) I'll add your code to GHC's
regression suite, so that all future releases will guarantee to run your
code.  [Or at least if not it'll be deliberate!]  Tell me if you'd like
this; it'll become visible in GHC's CVS repository.



More information about the Haskell mailing list