Gentle ghc maintainers:

I am trying to compile a medium-sized (~ 10,000 line) Haskell
program using ghc 5.02 on Windows 2000.  This program is 
generated automatically by a Java program, and contains 
Lambada-style type-safe bindings for allowing Haskell 
program to call Java code.

I have sucesfully compiled this file using ghc 5.02 under Linux.  (It
takes about 5 minutes of wall clock time on a 500 MHz Pentium II,
which is...impressive...but it does at least compile succesfully).

Under Windows 2000, this doesn't compile.  The ghc process grows to
over 70 MB in size (!), my system eventually runs low on VM, and then,
finally, ghc crashes.  I get a dialog box telling me that ghc performed
an illegal memory access at address 0x00000038, which leads me to
think ghc simply grew until malloc() failed, and then ignored the
return code from malloc.  I increased my swap file size, and that
seems only to have changed the error behavior slightly.  Now I get the
message:

c:\[...]\ghc.exe: fatal error: scavenge: unimplemented/strange closure type 0 @
090033D8
make: *** [HavenJavaBindings.o] Error 254

instead of the dialog box.

To make it possible for you to reproduce the problem, I have created a
tar-file with a snapshot of my source tree (including some library
binaries), and placed it at:

           haskell.org:~antony/ghc-win32-bug.tar.gz.

To reproduce the problem (on a Windows 2000 system with ghc-5.02 and
Cygwin), simply extract that tar file in some directory and do:

$ cd haven/src/haskell
$ make -f Makefile.ghc
[...]

Please let me know if you have any difficulties downloading the file
or reproducing the problem, or any suggestions for workarounds.  

Thanks!

        -Antony Courtney,
        Yale University
        <[EMAIL PROTECTED]>
-- 
Antony Courtney  
Grad. Student, Dept. of Computer Science, Yale University
[EMAIL PROTECTED]          http://www.apocalypse.org/pub/u/antony

_______________________________________________
Glasgow-haskell-bugs mailing list
[EMAIL PROTECTED]
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs

Reply via email to