hi, i can't really argue with simon he is the real compiler expert. and i am not familiar with the inner workings of GHC, so my argument is even weaker. and in any case discussions on the haskell mailing list very rarely have any effect except for their entertainment value during boring meetings... so if you are sitting bored somewhere read on :-) in my defence, we do have recursive modules implemented in our front end, and it can do type checking (which is presumably where the difficulties with recursive modules arise). and implementing the recursive modules was definitely not the hardest part to get working. .hi-boot files indeed do the job. so do seq, the monomorphism restriction, defaulting and unsafePerformIO. it is just that they don't do the job very well. i am curious however, what is difficult about implementing recursive modules (that is, if it can be explained without getting into the technical details of GHC). bye iavor Simon Peyton-Jones wrote:
| there is no need to use such hacks. it is not dificult to add suport | for mutually recursive modules to an implementation directly. | unfortunatley none of the working haskell implementations support | recursive modules,
Simple in principle, not so simple in practice. If it was easy to make GHC do mutually recursive modules we'd do it right away. It's certainly possible but it's real work, and .hi-boot files, while clunky, do the job.
S
-- ================================================== | Iavor S. Diatchki, Ph.D. student | | Department of Computer Science and Engineering | | School of OGI at OHSU | | http://www.cse.ogi.edu/~diatchki | ==================================================