gmp-4.2.1 build problems

[Peter Breitenlohner]

HI,

I'd like to report two problems encountered in building gmp-4.2.1
on an i686-pc-gnu-linux system (actually an AMD64 processor running with
32Bit versions of the kernel and gcc).

(1) The file configfsf.guess properly detects this as i686-pc-gnu-linux,
but your config.guess wrapper subsequently changes that to
x86_64-pc-linux-gnu causing the build process to fail (ABI=64 with 32 bit
limbs).

Of course one can override this with "configure --build=i686-pc-gnu-linux"
but I consider the necessity to do so as a gross misfeature (aka bug).

(2) I noticed that "configure --enable-fat" also fails and found that this
problem is addressed by the patch x86-fat.diff on your web page.

Unfortunately this patch contains a typo. The modification for configure
reads:

*** 3555,3560 ****
       esac

!     # If the user asked for a fat build, override the path set above
       if test $enable_fat = yes; then
         extra_functions="$extra_functions fat fat_entry"
         path="x86/fat x86"
--- 3555,3570 ----		<<<<< WRONG
       esac

!     # If the user asked for a fat build, override the path and flags set above
       if test $enable_fat = yes; then
+       gcc_cflags_cpu=""
+       gcc_cflags_arch=""
         extra_functions="$extra_functions fat fat_entry"
         path="x86/fat x86"

resulting in an error (premature end of file). This part should read:

*** 3555,3560 ****
       esac

!     # If the user asked for a fat build, override the path set above
       if test $enable_fat = yes; then
         extra_functions="$extra_functions fat fat_entry"
         path="x86/fat x86"
--- 3555,3562 ----		<<<<< CORRECT
       esac

!     # If the user asked for a fat build, override the path and flags set above
       if test $enable_fat = yes; then
+       gcc_cflags_cpu=""
+       gcc_cflags_arch=""
         extra_functions="$extra_functions fat fat_entry"
         path="x86/fat x86"

regards
Peter Breitenlohner <peb at mppmu.mpg.de>