Take a look at the Makefile for gmp. Look for the line CFLAGS= and
then when going to configure mpfr use:
configure <whatever options you normally use> CFLAGS="<whatevers
listed in the make">. This should allow you to build and install
mpfr. Apparently, theres issues with it's configuration scripts.
This should allow you to get mpfr installed. Let me know if this
completly solves your issues or if you are still having problems
getting gcc to compile.
As far as my issue:
I've made a seperate build directory and am still having the same
issues. Not specifying any parameters gives me an error that gmp/mpfr
are not found... but I've built and installed them in the default
location. If you guys have any other suggestions I'm all ears. I
would perfer not to have to jump to an earlier build...
~Nick
On Jul 12, 2008, at 10:10 AM, Dan Allen wrote:
I too am having problems building an Intel gcc 4.3.1 on Mac OS X
10.5.4 on a standard iMac. There are many issues.
I grab the gmp-4.2.2 distribution, ./configure, make, and make
install it to /usr/local. Everything is great.
ISSUE #1:
I then grab the mpfr-2.3.1 distribution, ./configure it, and the ./
configure fails. It says that the ABIs do not match. config.guess
in the gmp build gives returns 'core2-apple-darwin9.4.0', while
config.guess in the mpfr build returns 'i386-apple-darwin9.4.0'.
Hmm. Well, core2 is not the same as i386, but they are in some
sense compatible. How does one fix this?
ISSUE #2:
There are a lot of places in the configure scripts where you see
rm: conftest.dSYM: is a directory
Is there an autoconf that fixes this? rm needs to be rm -rf on Mac
OS X. (Is this because of a case-insensitive file system?)
ISSUE #3:
I really wish that these extra dependencies were not there. I can
build gcc 4.2.5 without any problems, but requiring gmp and mpfr in
4.3+ with these hassles is a real pain. Why can't these be
integrated into gcc and put into gcclib so builds go smoother?
Thanks for listening.
Dan Allen
