HUGS> :t (== []) flip (==) [] :: Eq a => [a] -> Bool GHCI> :t (== []) forall a. (Eq [a]) => [a] -> Bool I think ghc is doing the right thing. -Levent.
Hi Levent, | HUGS> :t (== []) | flip (==) [] :: Eq a => [a] -> Bool | | | GHCI> :t (== []) | forall a. (Eq [a]) => [a] -> Bool | | I think ghc is doing the right thing. As far as Haskell 98 is concerned, Hugs is only doing what is required to produce a type whose syntax conforms to the language report. As far as the language report is concerned, I agree with you. (And I wrote a fairly big chunk of text in my old dissertation to explain why.) Indeed, if you run Hugs in -98 mode, then it too produces: Main> :t (==[]) flip (==) [] :: Eq [a] => [a] -> Bool Main> All the best, Mark
As far as the language report is concerned, I agree with you. (And I wrote a fairly big chunk of text in my old dissertation to explain why.)
I see. Section 7.1 of your thesis explains this in quite detail, and I'm surprised to see that you've even used the exact same example there.
Indeed, if you run Hugs in -98 mode, then it too produces:
Main> :t (==[]) flip (==) [] :: Eq [a] => [a] -> Bool
I run Hugs out of the CVS repository, and in both modes (i.e. both -98, and +98) it produces the type: Eq a => [a] -> Bool However, I can verify the behavior you've mentioned using the Feb'2000 release. Something must have (accidentally?) changed since then. -Levent.
On Thursday 13 June 2002 02:30 am, Levent Erkok wrote:
However, I can verify the behavior you've mentioned using the Feb'2000 release. Something must have (accidentally?) changed since then.
Try :set +T. The default setting for this flag ("print most general type") was recently changed to -T. There may be something on the mailing list in the last month that will explain what's up with this. --Jeff
Spot on. The default behaviour of :type has now been flipped back to what it has always been. --sigbjorn ----- Original Message ----- From: "Jeffrey R Lewis" <jeff@galois.com> To: "Levent Erkok" <erkok@cse.ogi.edu>; "Mark P Jones" <mpj@cse.ogi.edu> Cc: <hugs-bugs@haskell.org>; "Mark P. Jones" <mpj@cse.ogi.edu> Sent: Thursday, June 13, 2002 09:53 Subject: Re: too much simplification?
On Thursday 13 June 2002 02:30 am, Levent Erkok wrote:
However, I can verify the behavior you've mentioned using the Feb'2000 release. Something must have (accidentally?) changed since then.
Try :set +T. The default setting for this flag ("print most general type") was recently changed to -T. There may be something on the mailing list in the last month that will explain what's up with this.
--Jeff _______________________________________________ Hugs-Bugs mailing list Hugs-Bugs@haskell.org http://www.haskell.org/mailman/listinfo/hugs-bugs
participants (4)
-
Jeffrey R Lewis -
Levent Erkok -
Mark P Jones -
Sigbjorn Finne