On 30/01/12 22:06, François Bissey wrote:
On Mon, 30 Jan 2012 16:20:50 Guilherme Brondani Torri wrote:
Hello!
I jut tried to compile Sage 4.8 and 5.0.beta1 on a OS X 10.6.8 (Snow
Leopard) without success.
I just did:
export MAKE="make -j8"
make
Even after R fails the compilation goes a bit further and stops. Please
see in attach the and r-2.14.0 log. Also attached the environment
variables form install.log.
For some reason R cannot find the symbols, it seems that the symbol *is*
available on the system, but the Sage lib doesn't have it. See below the
inspection of the libraries R tries to reach.
guilherme$ nm /usr/local/lib/libgomp.1.dylib | grep get_address
U ___emutls_get_address
guilherme$ nm
/Users/guilherme/sage-5.0.beta1/local/lib//libgcc_s.1.dylib | grep
get_address
<nothing>
Symbol is in another library:
guilherme$ nm /usr/local/lib/libgcc_s.1.dylib | grep get_address
0000000000011450 T ___emutls_get_address
I hope is not my system that is causing the failure.
Any ideas?
Looking at your R build log I'd say you have an issue with LD_LIBRARY_PATH.
You are after a symbol used by libgomp which is implementing openmp feature in
gcc, I am guessing you have installed some stuff in /usr/local (a gcc?).
In any case I am guessing this symbol should be provided by
/usr/local/libcc_s.1.dylib or possibly something else provided by xcode. But
because LD_LIBRARY_PATH doesn't include this location the linker is searching
in $SAGE_LOCAL/lib.
So what does your LD_LIBRARY_PATH read at this point and do you have
/usr/local/lib/libgcc_s.1.dylib ?
Francois
Thank you François!
Indeed I had a forgotten gcc compilation with the wrong prefix...
I cleaned it up and now Sage compiles as expected.
Sorry for the noise...
Guilherme
--
To post to this group, send an email to sage-devel@googlegroups.com
To unsubscribe from this group, send an email to
sage-devel+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
