Hello,
after about 12 hours of work I get the following:
[EMAIL PROTECTED] sage-2.8.3$ uname -a
SunOS neron 5.9 Generic_112233-08 sun4u sparc SUNW,Sun-Fire-480R
Solaris
[EMAIL PROTECTED] sage-2.8.3$ ./sage
----------------------------------------------------------------------
| SAGE Version 2.8.3, Release Date: 2007-08-31 |
| Type notebook() for the GUI, and license() for information. |
----------------------------------------------------------------------
sage: 1+1
2
sage:
Exiting SAGE (CPU time 0m0.01s, Wall time 0m4.26s).
We aren't 100% there yet, because I had to manually intervene in a
number of packages (usually ssue with linker flags, but that is a
build chain problem and specific to my setup). I build are more or
less complete userspace + development toolchain and I needed two
workarounds to compile libsage and satisfy the linker at startup.
All but 5 packages build (with minimal intervention for the reason
stated above) and those who do not build have either solutions that I
was too lazy to merge in or have temporary workarounds. Skipped
packages:
clisp-2.41.p8 (sbcl is an alternative, binary clisp 2.39 available for
Solaris)
maxima-5.13.0 (works fine with sbcl on Solaris, binary clisp)
genus2reduction-0.3
lcalc-20070107 (workaround exists)
cvxopt-0.8.2.p1 (complex.h issue)
But there are still some problems:
- GAP segfaults in the doctest, no idea why yet. GAP itself starts up
- libSingular segfault, I hope it is the strdup issue which we already
have a fix for, libSingular's doctest started with -gdb segfaults gdb :
(
- Singular's doctest hangs with python using 100% of one CPU and
Singular ideling, maybe an interface issue?
- DSage will probably fail due to missing proc file system, but we can
use local workarounds
- "partitions.so: symbol _Z4cosle: referenced symbol not found" - I
disabled importing that library for now. I have no clue what could be
wrong yet. The symbol is certainly not in an external library and
google draws a blank.
I am installing clisp from a binary now and will then build Maxima and
also try to fix the other packages. I assume that I will spend some
time during Sage's Bug Day 2 to debug some of the remainng issues.
William, feel free to make a copy /extra/home/mabshoff/SAGE-build/
sage-2.8.3 on neron and play around with it. Make sure though that you
do a
export LD_LIBRARY_PATH=/extra/home/mabshoff/tools/lib/:/extra/home/
mabshoff/tools/gcc-4.2.1/lib/:$LD_LIBRARY_PATH
first, though. I am sure that there are still enough bugs in the port
for everybody involved.
Cheers,
Michael
################################
Detailed remarks:
libgcrypt-1.2.4, opencdk-0.5.9m, gnutls-1.6.3, various other
packages:
The following problems might be due to too old/too recent libtool (not
sure). The problem boils down to that the Sun ld cannot handle any of
the options below (at least not in that form). A solution might be to
compile gcc with explicit Sun ld support.
# Flag to hardcode $libdir into a binary during linking.
# This must work even if $libdir does not exist.
hardcode_libdir_flag_spec="\${wl}--rpath \${wl}\$libdir"
remove '"\${wl}--rpath \${wl}\$libdir"'
# Commands used to build and install a shared archive.
archive_cmds="\$CC -shared -nostdlib \$predep_objects \$libobjs \
$deplibs \$postdep_objects \$compiler_flags \${wl}-soname \
$wl\$soname -o \$lib"
remove '\${wl}-soname \$wl\$soname'
# Compiler flag to generate shared objects directly from archives.
whole_archive_flag_spec="\${wl}--whole-archive\$convenience \${wl}--no-
whole-archiv
remove '"\${wl}--whole-archive\$convenience \${wl}--no-whole-archiv'
# Commands used to build and install a shared archive.
archive_cmds="\$CC -shared \$libobjs \$deplibs \$compiler_flags -o \
$lib"
# Commands used to build and install a shared archive.
archive_cmds="\$CC -shared \$libobjs \$deplibs \$compiler_flags -o \
$lib"
------------------
gsl fails to build, but make doesn't detect the problem, problem is
related to linker flags.
------------------
LinBox doesn't find gsl's BLAS problem, but that might be related to
the initial gsl failure. I Uses ATLAS anyway ;)
fortran-20070816 fails, but use experimental gcc.spkg with gfortran,
In sage_fortran refer to the right binary to make all Fortran packages
compile
numpy fails to link against ATLAS: order of arguments? But in the
second round
picks up other BLAS
----------------------
In scons-0.97/SCons/Tool/ edit __init__.py to prefer g++ & gcc over
Sun's CC.
To do that search for "elif str(platform) == 'sunos':" and swap the
order of gcc/g++ and the Sun compilers
------------------------------
sagelib: linux-ism in sage/modules/vector_modn_dense.pyx
gcc -fno-strict-aliasing -DNDEBUG -g -O3 -Wall -Wstrict-prototypes -
fPIC -I/extra/home/mabshoff/SAGE-build/sage-2.8.3/local//include -I/
extra/home/mabshoff/SAGE-build/sage-2.8.3/local//include/python -I/
extra/home/mabshoff/SAGE-build/sage-2.8.3/devel//sage/sage/ext -I/
extra/home/mabshoff/SAGE-build/sage-2.8.3/local/include/python2.5 -c
sage/modules/vector_modn_dense.c -o build/temp.solaris-2.9-sun4u-2.5/
sage/modules/vector_modn_dense.o -w
sage/modules/vector_modn_dense.c:33:20: error: stdint.h: No such file
or directory
sage/modules/vector_modn_dense.c:516: error: expected specifier-
qualifier-list before 'int_fast32_t'
sage/modules/vector_modn_dense.c:558: error: expected specifier-
qualifier-list before 'int_fast32_t'
sage/modules/vector_modn_dense.c:563: error: expected declaration
specifiers or '...' before 'int_fast32_t'
sage/modules/vector_modn_dense.c:564: error: expected specifier-
qualifier-list before 'int_fast32_t'
sage/modules/vector_modn_dense.c:572: error: expected specifier-
qualifier-list before 'int_fast64_t'
sage/modules/vector_modn_dense.c:577: error: expected declaration
specifiers or '...' before 'int_fast64_t'
sage/modules/vector_modn_dense.c:578: error: expected specifier-
qualifier-list before 'int_fast64_t'
We have a special stdint.h for Solaris, but I assume that didn't make
it into sagelib yet.
-----------------------------------
<type 'exceptions.ImportError'>: ld.so.1: sage.bin: fatal: relocation
error: file /extra/home/mabshoff/SAGE-build/sage-2.8.3/local/lib/
python2.5/site-packages/sage/rings/real_rqdf.so: symbol isinf:
referenced symbol not found
-----------------------------------
heilbronn.pyx: needs roundf workaround
#define roundf(a) ((fmod(a,1)<0.5)?floor(a):ceil(a))
--~--~---------~--~----~------------~-------~--~----~
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/
-~----------~----~----~----~------~----~------~--~---
