Possible issue with Hoogle and Haddock?
Hi, I was noticing recently that there seems to be a problem with Hoogle and Haddock. In particular, I just hoogled "bracket" and got the following result: bracket :: IO a -> a -> IO b -> a -> IO c -> IO c Clearly this is the wrong type, as it should be bracket :: IO a -> (a -> IO b) -> (a -> IO c) -> IO c The Haddock-generated documentation at least has newlines to group the terms, but this is still far from correct. I noticed the same thing a while back in Parsec's documentation for token, token :: Stream s Identity t => t -> String -> t -> SourcePos -> t -> Maybe a -> Parsec s u a which again should have parens. Is this a bug? Is is something that's well-known? Cheers, steve
On Thu, Nov 20, 2008 at 02:25:25PM -0500, Stephen Hicks wrote:
I was noticing recently that there seems to be a problem with Hoogle and Haddock. In particular, I just hoogled "bracket" and got the following result: bracket :: IO a -> a -> IO b -> a -> IO c -> IO c Clearly this is the wrong type, as it should be bracket :: IO a -> (a -> IO b) -> (a -> IO c) -> IO c
Is this a bug? Is is something that's well-known?
Yes, it's a known bug: http://hackage.haskell.org/trac/ghc/ticket/2584 Thanks Ian
I was noticing recently that there seems to be a problem with Hoogle and Haddock. In particular, I just hoogled "bracket" and got the following result: bracket :: IO a -> a -> IO b -> a -> IO c -> IO c Clearly this is the wrong type, as it should be bracket :: IO a -> (a -> IO b) -> (a -> IO c) -> IO c
Is this a bug? Is is something that's well-known?
Yes, it's a known bug: http://hackage.haskell.org/trac/ghc/ticket/2584
I can make a work around for Hoogle (well, I can write a workaround in the Hoogle specific code of Haddock), but was hoping that the bug would be fixed before 6.10. I guess its now worth making the Hoogle specific fix. Thanks Neil ============================================================================== Please access the attached hyperlink for an important electronic communications disclaimer: http://www.credit-suisse.com/legal/en/disclaimer_email_ib.html ==============================================================================
Hi Stephen,
I've now worked around this bug in Hoogle - I'm just about to rebuild
the website, and hopefully the bug will have disappeared. (Rebuilding
the website could take a few days, as I'm currently hunting for the
right compiler etc...)
Thanks
Neil
On Fri, Nov 21, 2008 at 8:21 AM, Mitchell, Neil
I was noticing recently that there seems to be a problem with Hoogle and Haddock. In particular, I just hoogled "bracket" and got the following result: bracket :: IO a -> a -> IO b -> a -> IO c -> IO c Clearly this is the wrong type, as it should be bracket :: IO a -> (a -> IO b) -> (a -> IO c) -> IO c
Is this a bug? Is is something that's well-known?
Yes, it's a known bug: http://hackage.haskell.org/trac/ghc/ticket/2584
I can make a work around for Hoogle (well, I can write a workaround in the Hoogle specific code of Haddock), but was hoping that the bug would be fixed before 6.10. I guess its now worth making the Hoogle specific fix.
Thanks
Neil
============================================================================== Please access the attached hyperlink for an important electronic communications disclaimer:
http://www.credit-suisse.com/legal/en/disclaimer_email_ib.html ==============================================================================
participants (4)
-
Ian Lynagh -
Mitchell, Neil -
Neil Mitchell -
Stephen Hicks