December 2001 - Nhc-bugs

CVS copy doesn't compile
by Ian Lynagh 18 Dec '01

18 Dec '01

I checked out the CVS tree an hour or so ago (at a guess) and make all (after ./configure with a --prefix) failed with: ----- /home/ian/cvs/nhc/script/nhc98 -cpp -c +CTS -part -redefine -CTS -P.. -P../PreludeIO -P../IO -prelude -o /home/ian/cvs/nhc/targets/ix86-Linux/obj/prelude/LowB/PrimOpenSocket.o PrimOpenSocket.hs Fail: GcodeFix.hs:165: Non-exhaustive patterns in function nthcon make[2]: *** [/home/ian/cvs/nhc/targets/ix86-Linux/obj/prelude/LowB/PrimOpenSocket.o] Error 1 make[2]: Leaving directory `/home/ian/cvs/nhc/src/prelude/LowB' make[1]: *** [LowB.make] Error 2 make[1]: Leaving directory `/home/ian/cvs/nhc/src/prelude' make: *** [targets/ix86-Linux/prelude] Error 2 ----- I uncommented the [] case and got Fail: nthcon: n==0 con==35 Thanks Ian

2 1

hmake builds the wrong thing
by Ian Lynagh 17 Dec '01

17 Dec '01

This one had me puzzled for a few seconds before I spotted what was going on! $ rm foo*; touch foo.hs foo.lhs; hmake foo.lhs nhc98 -c -o foo.o foo.hs Warning: Can not find main in module Main. Thanks Ian

2 3

Incorrect type clash
by Ian Lynagh 17 Dec '01

17 Dec '01

With this module: data Foo = Foo Int Bar deriving Show data Bar = Bar Int Int deriving Show a `Foo` (b `Bar` c) = 3 `Foo` (4 `Bar` 5) I get Error after type deriving/checking: Type error type clash between Prelude.Int and Main.Foo when trying to apply function at 5:15 to its 1:st argument at 5:6. while GHC and hugs are happy. Thanks Ian

2 3

"import qualified IO (readFile)" gives IO.readFile not defined
by Ian Lynagh 17 Dec '01

17 Dec '01

Hi guys, With import qualified IO (readFile) I get Error when renaming:: Identifier IO.readFile used at 2:22 is not defined. I vaguely remember seeing something like this on one of the lists recently, but I can't find it now, so my apologies if this has already been reported. Thanks Ian

2 1

Not able to complete compilation of nhc98 1.10
by Scott Williams 11 Dec '01

11 Dec '01

For some reason the nhc98 build isn't completing on my machine. Since the build listing is quite long, I've only included the tail. $ uname -a Linux scotts 2.4.16 #1 Wed Dec 5 18:24:14 PST 2001 i586 unknown $ ghc --help Use of the Glasgow Haskell Compiler driver: ghc-5.02.1 [command-line-options-and-input-files] .... The following patches were applied: patch-1.10-Array patch-1.10-arrow patch-1.10-blockbuffer patch-1.10-counter patch-1.10-ghc5.02 patch-1.10-hiding patch-1.10-keywords ... ... gmake[1]: Entering directory `/home/scott/Haskell/nhc98-1.10/src/tracer/hat' mkdir -p /home/scott/Haskell/nhc98-1.10/targets/ix86-Linux/obj/hat gcc -O3 -I/home/scott/Haskell/nhc98-1.10/include -DVERSION="\"1.10\"" -c nodecache.c -o /home/scott/Haskell/nhc98-1.10/targets/ix86-Linux/obj/hat/nodecache.o gcc -O3 -I/home/scott/Haskell/nhc98-1.10/include -DVERSION="\"1.10\"" -c minisocklib.c -o /home/scott/Haskell/nhc98-1.10/targets/ix86-Linux/obj/hat/minisocklib.o gcc -O3 -I/home/scott/Haskell/nhc98-1.10/include -DVERSION="\"1.10\"" -c utils.c -o /home/scott/Haskell/nhc98-1.10/targets/ix86-Linux/obj/hat/utils.o gcc -O3 -I/home/scott/Haskell/nhc98-1.10/include -DVERSION="\"1.10\"" -c browsercomms.c -o /home/scott/Haskell/nhc98-1.10/targets/ix86-Linux/obj/hat/browsercomms.o gcc -O3 -I/home/scott/Haskell/nhc98-1.10/include hat-stack.c /home/scott/Haskell/nhc98-1.10/targets/ix86-Linux/obj/hat/utils.o \ -o /home/scott/Haskell/nhc98-1.10/lib/ix86-Linux/hat-stack gcc -O3 -I/home/scott/Haskell/nhc98-1.10/include -DPRELSRCDIR=\"/home/scott/Haskell/nhc98-1.10/src/prelude\" hat-connect.c \ -o /home/scott/Haskell/nhc98-1.10/lib/ix86-Linux/hat-connect /home/scott/Haskell/nhc98-1.10/targets/ix86-Linux/obj/hat/nodecache.o /home/scott/Haskell/nhc98-1.10/targets/ix86-Linux/obj/hat/minisocklib.o /home/scott/Haskell/nhc98-1.10/targets/ix86-Linux/obj/hat/utils.o /home/scott/Haskell/nhc98-1.10/targets/ix86-Linux/obj/hat/browsercomms.o gcc -c -g -DVERSION="\"1.10\"" -o /home/scott/Haskell/nhc98-1.10/targets/ix86-Linux/obj/hat/nodelist.o nodelist.c gcc -c -g -DVERSION="\"1.10\"" -o /home/scott/Haskell/nhc98-1.10/targets/ix86-Linux/obj/hat/hatinterface.o hatinterface.c gcc -c -g -DVERSION="\"1.10\"" -o /home/scott/Haskell/nhc98-1.10/targets/ix86-Linux/obj/hat/hashtable.o hashtable.c gcc -c -g -DVERSION="\"1.10\"" -o /home/scott/Haskell/nhc98-1.10/targets/ix86-Linux/obj/hat/observe.o observe.c gcc -c -g -DVERSION="\"1.10\"" -o /home/scott/Haskell/nhc98-1.10/targets/ix86-Linux/obj/hat/detect.o detect.c gcc -c -g -DVERSION="\"1.10\"" -o /home/scott/Haskell/nhc98-1.10/targets/ix86-Linux/obj/hat/Expressions.o Expressions.c gcc -c -g -DVERSION="\"1.10\"" -o /home/scott/Haskell/nhc98-1.10/targets/ix86-Linux/obj/hat/hatgeneral.o hatgeneral.c gcc -c -g -DVERSION="\"1.10\"" -o /home/scott/Haskell/nhc98-1.10/targets/ix86-Linux/obj/hat/FunTable.o FunTable.c ar -rc /home/scott/Haskell/nhc98-1.10/targets/ix86-Linux/obj/hat/hatlib.a /home/scott/Haskell/nhc98-1.10/targets/ix86-Linux/obj/hat/detect.o /home/scott/Haskell/nhc98-1.10/targets/ix86-Linux/obj/hat/nodelist.o /home/scott/Haskell/nhc98-1.10/targets/ix86-Linux/obj/hat/hatinterface.o /home/scott/Haskell/nhc98-1.10/targets/ix86-Linux/obj/hat/FunTable.o /home/scott/Haskell/nhc98-1.10/targets/ix86-Linux/obj/hat/Expressions.o /home/scott/Haskell/nhc98-1.10/targets/ix86-Linux/obj/hat/hashtable.o /home/scott/Haskell/nhc98-1.10/targets/ix86-Linux/obj/hat/observe.o /home/scott/Haskell/nhc98-1.10/targets/ix86-Linux/obj/hat/hatgeneral.o /home/scott/Haskell/nhc98-1.10/script/hmake -nhc98 -K10M +CTS -H20M -CTS Hatobserve /home/scott/Haskell/nhc98-1.10/targets/ix86-Linux/obj/hat/hatlib.a -d/home/scott/Haskell/nhc98-1.10/targets/ix86-Linux/obj/hat hmake: Sorry, I do not know about your installation of nhc98 gmake[1]: *** [/home/scott/Haskell/nhc98-1.10/lib/ix86-Linux/hat-observe] Error 1 gmake[1]: Leaving directory `/home/scott/Haskell/nhc98-1.10/src/tracer/hat' gmake: *** [lib/ix86-Linux/hat-stack] Error 2

2 1

Parse problems
by Ian Lynagh 11 Dec '01

11 Dec '01

Hi, With the module { main = undefined foo = 5 } I get 4:6 Found = but expected one of }' ; {-end-of-definition-} prefix _infix_ _infixr_ _infixl_ <string> <char> <rational> <integer> ~ [ ( _ <conid> <varid> ` <varop> <conop> foreign _default_ _instance_ _class_ _data_ _newtype_ _type_ and with module Foo where { foo = 5 bar = 6 } I get 4:6 Found = but expected one of }' ; {-end-of-definition-} prefix _infix_ _infixr_ _infixl_ <string> <char> <rational> <integer> ~ [ ( _ <conid> <varid> ` <varop> <conop> foreign _default_ _instance_ _class_ _data_ _newtype_ _type_ Thanks Ian

2 2

the format of error diagnostics
by Feliks Kluzniak 10 Dec '01

10 Dec '01

Hello, nhc98 would be much more convenient to use if the error diagnostics were compatible with Emacs compile mode. In other words, instead of nhc98 -c -d ../nhc Flatten.hs ==================================== Error when renaming:: Constructor OK used at 833:6 is not defined. it would be nice to have the message appear as Flatten.hs:833:6: Error when renaming:: Constructor OK is not defined. Regards, -- F.

2 4

--+ not treated as a start of comment
by Ian Lynagh 09 Dec '01

09 Dec '01

If I have foo = 0 --+ 1 then nhc tells me Identifier --+ used at 2:11 is not defined. while the report (and hugs) believe "--+ 1" is a comment. Thanks Ian

1 0

export and import list (,) not allowed
by Ian Lynagh 09 Dec '01

09 Dec '01

With the module module Foo (,) where foo = 0 I get 2:14 Found Prelude.2 but expected one of _where_ () ( but the report (and GHC) allows this. Similarly with import lists I get either 4:15 Found Prelude.2 but expected a {-end-of-definition-or-EOF-} or 4:15 Found _hiding_ but expected a {-end-of-definition-or-EOF-} depending on whether I use hiding. Thanks Ian

1 0

Re: [nhc-bugs] problems compiling
by Malcolm Wallace 04 Dec '01

04 Dec '01

> > Problem 2, when trying to compile for tracing: > > module T1 where > instance (Enum a, Bounded a, Eq b) => Eq (a -> b) where > f1 == f2 = and [f1 x == f2 x | x <- [minBound .. maxBound]] > > > Fail: The class Prelude.Enum has no instance for the type Prelude.R. > > Possible sources for the problem are: 9:6 Ok, the problem here is not so much a bug but more of a restriction on how the tracing mechanism works. There is a workaround, although I'm afraid it may not be suitable for your application - declare a newtype or data for the function arrow, e.g. module T1 where newtype Func a b = Func (a->b) instance (Enum a, Bounded a, Eq b) => Eq (Func a b) where (Func f1) == (Func f2) = and [f1 x == f2 x | x <- [minBound .. maxBound]] Then you will need to wrap each function in a Func constructor before doing the equality comparison. Regards, Malcolm

1 0