I've used it in two classes now (the current one a "real" compiler
course), and have had very few problems. The Emacs-based user
interface is especially nice -- it supports separate compilation,
allows the user to turn compiler diagnostics on and off, will run
"Dialogues" as well as evaluate arbitrary expressions, and has a
mechanism for creating local "extensions" to modules without causing
recompilation of the module. Both the compiler and run-time
performance certainly have room for improvement, but nevertheless I
highly recommend Yale Haskell for anyone having a SparcStation with
lots (:-) of memory. Kudos to John Peterson and his co-workers
(which, by the way, does not include me) at Yale for putting this
system together and making it available for public consumption.
At the risk of getting into hot water with John Peterson, permit me
to put in a plug for the latest release of the Yale Haskell System:
It is a very robust and pleasant-to-use implementation of Haskell!
I don't want to contribute to a whole host of kudos to the Yale
Haskell system, but I wanted to add my own opinion as one separate
from Yale University and the community there. I have also found the
Yale Haskell system pleasent to use, and fairly robust. More
importantly, I was able to use the thing without constant support from
the Yale team. The level of documentation supplied is not fulsome,
but it is sufficient for the serious user. The Emacs interface is
very nice indeed.
This is not to cast aspersions on any other implementation; Gofer is
also interesting, for different reasons; other implementations also
have their points. I just wanted to second Dr. Hudak's words with
experience from someone outside the Yale community.