didier deshommes wrote: > > 2007/9/25, mabshoff <[EMAIL PROTECTED]>: >> >> >> >> On Sep 24, 10:42 pm, "William Stein" <[EMAIL PROTECTED]> wrote: >> > On 9/24/07, didier deshommes <[EMAIL PROTECTED]> wrote: >> > >> > > Some build notes: >> > > - Singular needs "-lcurses" whenever "-lreadline" is specified. >> Using >> > > it from the console still fails for me. >> >> It depends how readline is build. Many times -ltermcap should fix the >> problem, but I got Singular to build and run without problems on >> Solaris 9/10 Sparc and Solaris 10 Opteron. Another solution might be >> to build ncurses, but I would have to look at the beautiful and well >> structured Singular build system 8) > > > I'm able to start it from the console (with -ltermcap), but I had to > change the singular.py interface so that the executable would point to > $SAGE_LOCAL/bin/Singular-3-0-3 instead of SAGELOCAL/bin/Singular. I'm > guessing a link didn't get updated or something... > > >> I think the trick to building clisp on Solaris is to use a gcc 3.4 and >> pass -DNOGENERATIONALGC or something during the build. That is at >> least the workaround on NetBSD and when I build Sage's clisp on >> Solaris 9 the last time it crashes during the first invocation of the >> garbage collector. I reported the problem to the clisp development >> list but never got a reply. > > Painful.
Indeed, but as my understand goes William has some interested party for Sage on Solaris that wants to compile every bit from scratch. > >> >> > > - to build matplotlib, python needs this fix: >> > >http://www.scipy.org/Cookbook/Matplotlib/CompilingMatPlotLibOnSolaris... >> >> Interesting, never had that problem. > > Very interesting, I've had to build it on Solaris and nexenta and each > time compilation of matplotlib failed until I tried this. > >> Easiest is to symbolically link gfortran or g95 to g77, otherwise > > I did that, and I get this: > """ > customize GnuFCompiler > Couldn't match compiler version for 'G95 (GCC 4.0.3 (g95 0.90!) Aug 1 > 2006)\nCopyright (C) 2002-2005 Free Software Foundation, Inc.\n\nG95 > comes with NO WARRANTY, to the extent permitted by law.\nYou may > redistribute copies of G95\nunder the terms of the GNU General Public > License.\nFor more information about these matters, see the file named > COPYING' > customize Gnu95FCompiler > customize SunFCompiler > 'linker_exe' > [...] > > f90:f77: Lib/fftpack/dfftpack/dcosqb.f > sh: f90: not found > sh: f90: not found > error: Command "f90 -ftrap=%none -fixed -xcode=pic32 -c -c > Lib/fftpack/dfftpack/dcosqb.f -o > build/temp.solaris-2.11-i86pc-2.5/Lib/fftpack/dfftpack/dcosqb.o" > failed with exit status 1 > Error building scipy. > """ > > Do you know how scipy picks its compiler? The build scripts look opaque to > me... > That put it quite nicely, I looked at it for a while and figured that somebody else should try fixing that first. > Testing of interfaces is completely broken for me, I can't tell if I > miscompiled pexpect or if it's something much scarier/deeper. > I don't think pexpect is miscompiled on Solaris, but my Singular script in $SAGE_LOCAL/bin was screwed up somehow. I fixed that manually and didn't investigate what the problem was. Re pexpect vs. Solaris: All the other interfaces work, so it might be an oddity with Singular in particular. Even 1+1 in Sage via Singular would fail. > > didier > Cheers, Michael >> > > Tests that failed: >> > >> > Yikes! >> > >> > >> > >> > > sage -t algebras/free_algebra_quotient.py >> > > sage -t calculus/calculus.py >> > > sage -t ext/interactive_constructors_c.pyx >> > > sage -t functions/constants.py >> > > sage -t functions/functions.py >> > > sage -t functions/transcendental.py >> > > sage -t geometry/lattice_polytope.py >> > > sage -t interfaces/gap.py >> > > sage -t interfaces/singular.py >> > > sage -t lfunctions/lcalc.py >> >> I got a fix for lcalc. Need to send it in :) >> >> > > sage -t lfunctions/sympow.py >> >> No fix for Sympow on Solaris with Intel/AMD chips so far, but I >> believe that the problem is that sympow detects an Intel/AMD cpu and >> uses special assembly flags to use the extended FPU precision, while >> the fallback mode should be used. >> >> > > sage -t matrix/matrix2.pyx >> > > sage -t matrix/matrix_integer_dense.pyx >> > > sage -t matrix/matrix_mpolynomial_dense.pyx >> > > sage -t matrix/matrix_space.py >> > > sage -t modular/ssmod/ssmod.py >> > > sage -t plot/plot.py >> > > sage -t rings/number_field/number_field.py >> > > sage -t rings/number_field/number_field_element.pyx >> > > sage -t rings/number_field/number_field_ideal_rel.py >> > > sage -t rings/polynomial/groebner_fan.py >> > > sage -t rings/polynomial/multi_polynomial_element.py >> > > sage -t rings/polynomial/multi_polynomial_ideal.py >> > > sage -t >> rings/polynomial/multi_polynomial_ideal_libsingular.pyx >> > > sage -t rings/polynomial/multi_polynomial_libsingular.pyx >> > > sage -t rings/polynomial/multi_polynomial_ring.py >> > > sage -t rings/polynomial/multi_polynomial_ring_generic.pyx >> > > sage -t rings/polynomial/polynomial_element.pyx >> > > sage -t rings/polynomial/polynomial_quotient_ring.py >> > > sage -t >> rings/polynomial/polynomial_quotient_ring_element.py >> > > sage -t rings/polynomial/polynomial_singular_interface.py >> > > sage -t rings/polynomial/term_order.py >> > > sage -t rings/polynomial/toy_buchberger.py >> > > sage -t rings/complex_double.pyx >> > > sage -t rings/homset.py >> > > sage -t rings/morphism.py >> > > sage -t rings/power_series_ring.py >> > > sage -t rings/quotient_ring.py >> > > sage -t rings/quotient_ring_element.py >> > > sage -t rings/real_double.pyx >> > > sage -t rings/real_rqdf.pyx >> > > sage -t rings/ring.pyx >> > > sage -t schemes/elliptic_curves/ell_padic_field.py >> > > sage -t schemes/elliptic_curves/ell_rational_field.py >> > > sage -t schemes/generic/affine_space.py >> > > sage -t schemes/generic/algebraic_scheme.py >> > > sage -t schemes/generic/divisor.py >> > > sage -t schemes/generic/morphism.py >> > > sage -t schemes/generic/projective_space.py >> > > sage -t >> schemes/hyperelliptic_curves/hyperelliptic_padic_field.py >> > > sage -t schemes/plane_curves/affine_curve.py >> > > sage -t schemes/plane_curves/constructor.py >> > > sage -t schemes/plane_curves/curve.py >> > > sage -t schemes/plane_curves/projective_curve.py >> > > sage -t structure/element.pyx >> > >> >> Overall I am down to 17 failed tests and I believe that at least some >> of them might be related to the size of various int types. I will >> investigate those down the road as I come back from SD5, but if >> anybody else wants to try feel free to send in patches ;) >> >> > > > didier >> > >> > > > > Cheers, >> > >> > > > > Michael >> > >> >> Cheers, >> >> Michael >> >> > -- >> > William Stein >> > Associate Professor of Mathematics >> > University of Washingtonhttp://wstein.org >> >> >> > >> > > > > --~--~---------~--~----~------------~-------~--~----~ To post to this group, send email to sage-devel@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/sage-devel URLs: http://sage.scipy.org/sage/ and http://modular.math.washington.edu/sage/ -~----------~----~----~----~------~----~------~--~---
