Strangely after rebuilding under these conditions it is now failing on mpir (it is capable of building software so it isn't completely broken):
configure:4497: checking compiler /usr/local/bin/gcc -O2 -fomit-frame- pointer Test compile: configure:4511: /usr/local/bin/gcc -O2 -fomit-frame-pointer conftest.c >&5 libc.so.1: cc1: fatal: relocation error: file /usr/lib/libmpfr.so.1: symbol __gmpn_clz_tab: referenced symbol not found gcc: Internal error: Killed (program cc1) Please submit a full bug report. I'm not sure why it doesn't see or use libmpfr: bash-3.2$ ls /usr/local/lib/libmpfr.* /usr/local/lib/libmpfr.a /usr/local/lib/libmpfr.so /usr/local/lib/ libmpfr.so.1.2.0 /usr/local/lib/libmpfr.la /usr/local/lib/libmpfr.so.1 On Aug 5, 2:14 pm, "brandon.bar...@gmail.com" <brandon.bar...@gmail.com> wrote: > I did try setting LD_LIBRARY_PATH but this resulted in the same > problem. I even tried temporarily changing the ld.config search order > with crle (the solaris cousin of the linux ldconfig) from: > > /lib:/usr/lib:/usr/local/lib > > to: > /usr/local/lib:/lib:/usr/lib > > But this didn't help. Perhaps If I rebuild gcc with this, this > linking configuration ... > > Could you post your gcc configuration (gcc -v) that you use on your > SXCE machine whenever you have a chance to? > > Thanks, > > On Aug 5, 12:12 pm, "Dr. David Kirkby" <david.kir...@onetel.net> > wrote: > > > brandon.bar...@gmail.com wrote: > > > Actually this seems to be an issue with it trying to link to the > > > system mpfr. Shouldn't sage built its own mpfr? > > > I agree with you it looks like termcap is linking to the wrong library, > > but it does appear that gcc has generated an internal error. That is > > another issue altogether. > > > Looking at the output you showed some time back of gcc -v, I note > > > with-gmp-lib=/usr/local/lib --with-mpfr-include=/usr/local/include/ > > mpfr --with-mpfr-lib=/usr/local/lib > > > That means gcc was built with those libraries present. If those > > libraries are no longer present, or gcc is not finding them, then gcc > > will generate an internal error. That is what is happening below. > > > You could try adding /usr/local/lib to your LD_LIBRARY_PATH and seeing > > if that fixes it. > > > Until you have a compiler which does generate internal errors, you have > > no chance of Sage building. I suspect your compiler is not finding mpfr > > of gmp in /usr/local/lib > > > Dave > > > > On Aug 4, 9:01 pm, "brandon.bar...@gmail.com" > > > <brandon.bar...@gmail.com> wrote: > > >> Well, I didn't have any trouble with MPIR in 4.1.1.rc0! > > > >> bash-3.2$ ls spkg/installed/ > > >> bzip2-1.0.5 conway_polynomials-0.2 > > >> dir-0.1 mpir-1.2.p4 > > >> prereq-0.3 sage_scripts-4.1.1.rc0 > > > >> But it looks like termcap is having linker problems - I'll try using > > >> gnu ld/as toolchain > > > >> gmake[2]: Entering directory `/export/home/brandon/sage-4.1.1.rc0/spkg/ > > >> build/termcap-1.3.1.p0/src' > > >> /usr/local/bin/gcc -c -DTERMCAP_FILE=\"/etc/termcap\" -I. -I. -m32 - > > >> fPIC termcap.c > > >> ld.so.1: cc1: fatal: relocation error: file /usr/lib/libmpfr.so.1: > > >> symbol __gmpn_clz_tab: referenced symbol not found > > >> gcc: Internal error: Killed (program cc1) > > >> Please submit a full bug report. > > >> See <http://gcc.gnu.org/bugs.html> for instructions. > > >> gmake[2]: *** [termcap.o] Error 1 > > >> gmake[2]: Leaving directory `/export/home/brandon/sage-4.1.1.rc0/spkg/ > > >> build/termcap-1.3.1.p0/src' > > >> ERROR building termcap > > > >> On Aug 4, 8:45 pm, "brandon.bar...@gmail.com" > > > >> <brandon.bar...@gmail.com> wrote: > > >>> I think my current gcc toolchain may be doomed, as it uses the sun > > >>> linker and the gnu assembler! (I read this was a good thing to do, at > > >>> least at some point in history). > > >>> Configured with: ../gcc-4.4.0/configure --prefix=/usr/local --enable- > > >>> shared --disable-static --disable-libtool-lock --enable-objc-gc -- > > >>> enable-concept-checks --disable-libada --enable-libssp --enable- > > >>> languages=c,c++,objc,fortran --enable-threads=posix --enable-tls=yes -- > > >>> with-system-zlib --without-gnu-ld --with-ld=/usr/ccs/bin/ld --with-gnu- > > >>> as --with-as=/usr/sfw/bin/gas --with-gmp-include=/usr/include/gmp -- > > >>> with-gmp-lib=/usr/local/lib --with-mpfr-include=/usr/local/include/ > > >>> mpfr --with-mpfr-lib=/usr/local/lib --enable-c99 --enable-nls --enable- > > >>> wchar_t --enable-libstdcxx-allocator=mt --with-pic > > >>> I'll have a go anyway with the following and see what happens before > > >>> building a new toolchain: > > >>> bash-3.2$ export PATH=/usr/ccs/bin:/usr/sfw/bin:/usr/local/bin:$PATH > > >>> bash-3.2$ echo $PATH > > >>> /usr/ccs/bin:/usr/sfw/bin:/usr/local/bin:/opt/sunstudio12.1/bin:/opt/ > > >>> dtbld/bin:/export/home/brandon/bin:/opt/jdsbld/bin:/usr/gnu/bin:/usr/ > > >>> bin:/usr/ccs/bin:/usr/sbin:/bin:/usr/sfw/bin:/usr/X11/bin > > >>> I also had a problem building MPIR on OpenSolaris due to the linker > > >>> problems you noted on the wiki; if you can get mpir to use a correct > > >>> version of libtool (the one in /usr/bin worked) this problem > > >>> disappears. > > >>> On Aug 3, 8:29 pm, David Kirkby <david.kir...@onetel.net> wrote: > > >>>> 2009/8/3 brandon.bar...@gmail.com <brandon.bar...@gmail.com>: > > >>>>> I look forward to testing and building it on opensolaris / intel. > > >>>> Give Sage a try on OpenSolaris and let me know what happens. I posted > > >>>> ealier today some notes on what I needed to do to get it build on > > >>>> SPARC. OpenSolaris might be a bit different. Things might work or > > >>>> might not work. But some feedback would be useful. > > >>>> There is one thing to watch out for. Some of the files check the > > >>>> linker to see whether it is GNU or Sun and so give appropiate flags. > > >>>> What is actually needed is the linker that gcc uses. But my tests are > > >>>> based on the first linker in the path. So if your gcc was configured > > >>>> to use the Sun linker, make sure the location of the Sun linker > > >>>> (/usr/ccs/bin/) is in the path before any GNU linker from binutils. > > >>>> Conversely, if your gcc is configured to use the GNU linker and > > >>>> assembler, make sure the path to those from binutils is in the path > > >>>> before those of the Sun tools (/usr/ccs/bin). > > >>>> You can see what linker and assember gcc use, using 'gcc -v' > > >>>> I''ve not got OpsnSolaris at home but do haveSolarisExpress > > >>>> Community Edition on a laptop. MPIR refused to build properly on that, > > >>>> but others have got MPIR to work OK. > > >>>> Any feedback is welcome. --~--~---------~--~----~------------~-------~--~----~ 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 URLs: http://www.sagemath.org -~----------~----~----~----~------~----~------~--~---
