GHC 4.04 builds bad obj with GNU 2.95 on Solaris
gcc 2.95 seems to cause ghc 4.04 to build objects that cause segmentation faults. When I went back to gcc 2.7.2 everything worked fine. I also tried ghc 2.10 with gcc 2.95 and had NO problem. A simple [main = putStr "Hello\n"] demonstrates the effect. Something like module Main where import System main = do argv <- getArgs prog <- getProgName putStrLn (unwords (prog:argv)) exitWith ExitSuccess may show a bit more background info. Compiling it will cause the following to be displayed after the normal module version status line: still has jump to fast entry point: .proc 063120 c10l_entry: !#PROLOGUE# 0 save%sp, -112, %sp !#PROLOGUE# 1 add %i0, 4, %o0 cmp %o0, %i1 bleu,a .LL113 ld [%i0], %l2 callstg_update_PAP, 0 nop ld [%i0], %l2 .LL113: callc10l_fast1, 0 add %i0, 4, %i0 .LLfe5: still has jump to fast entry point: .proc 063120 c10q_entry: !#PROLOGUE# 0 save%sp, -112, %sp !#PROLOGUE# 1 add %i0, 4, %o0 cmp %o0, %i1 bleu,a .LL119 ld [%i0], %l2 callstg_update_PAP, 0 nop ld [%i0], %l2 .LL119: callc10q_fast1, 0 add %i0, 4, %i0 .LLfe7: still has jump to fast entry point: .proc 063120 c10s_entry: !#PROLOGUE# 0 save%sp, -112, %sp !#PROLOGUE# 1 add %i0, 4, %o0 cmp %o0, %i1 bleu,a .LL126 ld [%i0], %l2 callstg_update_PAP, 0 nop ld [%i0], %l2 .LL126: callc10s_fast1, 0 add %i0, 4, %i0 .LLfe9: Regards, Martin PS: I was trying to find a related problem report on the bugs mailing list archive. I find it hard to locate related info through trial and error guessing from the subject lines only. Is there a good way to search the complete data base for text strings?
Message repeats: Alleged WinTel v4.03 bugs?
I wrote: > Two oddities I've noticed with ghc-4.03, win+cygnus binary build: > > > The -ansi flag makes gcc go berserk on the generated code. I stopped. > > > 'getEnv "PATH"' returns, not the value of PATH, but the whole environment. > Which is also handy, but not what I understood the report to specify, > and not what ghc-4.02 does on this 'ere Sun. Can someone comment on (esp.) the second of these? Is this reproducible, is it the case _only_ on Win32, and, is it/will it be fixed in 4.04, or thereafter? (In the meantime, I guess a work-around is fairly obvious.) Cheers, Alex.
confusing error message
ghc-4.04 gives the following confusing error message: compiling with ghc-4.04 -fglasgow-exts -c Test.hs Test.hs:8: None of the type variable(s) in the constraint `Eq a' appears in the type `Set a -> Set a -> Set a' In the type signature for `unionSets' Compilation had errors module Test where unionSetB :: Eq a => Set a -> Set a -> Set a unionSetB (s1 :: Set a) s2 = unionSets s1 s2 where unionSets :: Eq a => Set a -> Set a -> Set a unionSets a b = a Ok, I know why this happens, because a is no longer free in unionSets because of the pattern signature, but initially it confused me looking at it trying to work out why a /= a. (Incidently hugs accepts this code without any complaints) Meurig
Bug in ghc/rts/Select.c
On Sparc-Solaris it is incorrect to call the select function with a negative time value. Nevertheless rts/Select.c does this. I seem to have fixed the problem by stopping it decrementing the delay field of TSO.block_info below zero at the end of the call. Attached is the fixed version. Select.c
downloading Haskell
Hi, Could you help me please; I'm looking to possibly be able to download a Haskell compiler in as simple a way as possible (preferably without zip files -- if possible -- or if so,together with instrucs on how to use zip files as I'm not very experienced in using it) --- preferably to be able to download to floppy disk initially and then to be able to transfer to my home computer. I did have a look at the Installation instrucs for the Glasgow Haskell compiler homepage but still seemed to be rather complicated. BE most grateful if you could possibly assist. Thanks