logo       

RE: GHC build break in Win32Spawn.hs: msg#00019

lang.haskell.glasgow.bugs

Subject: RE: GHC build break in Win32Spawn.hs

Are you building with GHC 6.2.2? This fails in GHC 6.2.2 because
openFd takes one less parameter than in 6.3.x (if I remember
correctly, this change was made by Simon M. about 7 months ago in
GHC.Handle). I was able to "make ; make install ; make stage3"
GHC-6.3.20050106, compiling with GHC 6.2.2 (as released) on MSYS, by
removing the last argument of each of the three to openFd in
Win32Spawn.hs, but that surely isn't the correct solution, because I
was able to build 6.3.20041223 using GHC 6.2.2 without any problem.

Note also that recent GHC 6.3.x builds fail when compiled with
(recent) 6.3.x versions because of the "Bug in libcompat" described by
Krasimir Angelov (Distribution.Compat.ReadP is visible using GHC
6.2.2, but it is hidden in its package in recent GHC 6.3.x.)

By the way, if you have any tips about how to build GHC faster on
Windows (or any pointers to pre-built nightly snapshots), I would love
to have them. Building GHC takes a very long time on my laptop (more
than two hours).

Regards,
Brian Smith

Has anyone lese noticed this build break when building GHC for Windows?

Win32Spawn.hs:38:14:
Couldn't match `IO Handle' against `t -> t1'
Expected type: IO Handle
Inferred type: t -> t1
Probable cause: `openFd' is applied to too many arguments in the call
(openFd (fromIntegral wIn)
(Just RegularFile)
False
("<fd " ++ ((show wIn) ++ ">"))
WriteMode
False
False)
In a 'do' expression:
hIn <- openFd (fromIntegral wIn)
(Just RegularFile)
False
("<fd " ++ ((show wIn) ++ ">"))
WriteMode
False
False


<Prev in Thread] Current Thread [Next in Thread>
Google Custom Search

News | FAQ | advertise