François Bissey
unread,Dec 16, 2012, 8:48:07 PM12/16/12Sign in to reply to author
Sign in to forward
You do not have permission to delete messages in this group
Either email addresses are anonymous for this group or you need the view member email addresses permission to view the original message
to lmnd-...@googlegroups.com
Hi all,
Just a general email to inform you of what is happening this side.
sage-5.4:
There has been one invasive change because we dropped
etc/env.d/99sage.
We now have a file sage/misc/variables.py which mostly
mirror sage-env. Now that this file exist sympy and mpmath
can be started without triggering errors about sage variables
not being defined. The only problem is if you have gmpy installed
as well, look in sage-env on how to deal with this.
This also means that sage can be imported from the python prompt....
Why did I need to make a separate file? Variables were defined in
sage/misc/misc.py and initially I tried to replace calls to os.environ
with imports from sage.misc.misc but this is not always possible
and breaks the building of sage. misc.py imports randstate which
is from a pyx file. So anything involved in the installation/compilation
involving sage/misc/misc.py fails.
Other stuff: I am starting to move things to numpy 1.7.0 to be released
soon. numpy 1.5.1 is a source of brokeness to a lot of people in Gentoo
who are using sage-on-gentoo, I am considering the move as urgent.
I am also looking with other sage developers to upgrade ecl/maxima to
something a bit more recent. That'll probably land in sage-on-gentoo before
sage proper.
OS X side: gcc-apple on prefix is showing its age. If you are using openblas
it now needs to be compiled with clang:
emerge -uDv clang
CC=clang emerge -uv openblas
There are issues with compiling linbox-1.3.2, gcc-apple won't do and neither
will clang (with different errors) I have to check upstream. I managed to
compile it by using an unmasked gcc-4.6.3. However you cannot switch
to this version of gcc to build and maintain the prefix it leads to a number
of problems unfortunately.
polybori refuses to build with boost 1.52 which is what I currently have
there:
i686-apple-darwin9-g++ -o libpolybori/src/CCuddFirstIter.os -c -O2 -pipe -
march=nocona -ggdb -fno-strict-aliasing -fno-strict-aliasing -std=c++98 -
ftemplate-depth-100 -mmmx -msse -msse2 -msse3 -fPIC -DPBORI_NDEBUG -
DSIZEOF_VOID_P=4 -DSIZEOF_INT=4 -DSIZEOF_LONG=4 -DPBORI_HAVE_HASH_MAP -
DPBORI_HAVE_M4RI -DPBORI_HAVE_M4RI_PNG -Ilibpolybori/include -
Igroebner/include -I/Users/frb15/Desktop/Gentoo/usr/include/python2.7
libpolybori/src/CCuddFirstIter.cc
/Users/frb15/Desktop/Gentoo/usr/include/boost/range/algorithm/equal.hpp: In
function 'void export_monomial()':
/Users/frb15/Desktop/Gentoo/usr/include/boost/range/algorithm/equal.hpp:160:
error: 'boost::range' is not a function,
/Users/frb15/Desktop/Gentoo/usr/include/boost/python/iterator.hpp:112: error:
conflict with 'template<class NextPolicies, class Target, class Accessor1,
class Accessor2> boost::python::api::object boost::python::range(Accessor1,
Accessor2, NextPolicies*, boost::type<U>*)'
PyPolyBoRi/monomial_wrapper.cc:66: error: in call to 'range'
i686-apple-darwin9-g++ -o libpolybori/src/BooleMonomial.os -c -O2 -pipe -
march=nocona -ggdb -fno-strict-aliasing -fno-strict-aliasing -std=c++98 -
ftemplate-depth-100 -mmmx -msse -msse2 -msse3 -fPIC -DPBORI_NDEBUG -
DSIZEOF_VOID_P=4 -DSIZEOF_INT=4 -DSIZEOF_LONG=4 -DPBORI_HAVE_HASH_MAP -
DPBORI_HAVE_M4RI -DPBORI_HAVE_M4RI_PNG -Ilibpolybori/include -
Igroebner/include -I/Users/frb15/Desktop/Gentoo/usr/include/python2.7
libpolybori/src/BooleMonomial.cc
i686-apple-darwin9-g++ -o groebner/src/polynomial_properties.os -c -O2 -pipe -
march=nocona -ggdb -fno-strict-aliasing -fno-strict-aliasing -std=c++98 -
ftemplate-depth-100 -mmmx -msse -msse2 -msse3 -fPIC -DPBORI_NDEBUG -
DSIZEOF_VOID_P=4 -DSIZEOF_INT=4 -DSIZEOF_LONG=4 -DPBORI_HAVE_HASH_MAP -
DPBORI_HAVE_M4RI -DPBORI_HAVE_M4RI_PNG -Ilibpolybori/include -
Igroebner/include -I/Users/frb15/Desktop/Gentoo/usr/include/python2.7
groebner/src/polynomial_properties.cc
scons: *** [PyPolyBoRi/monomial_wrapper.os] Error 1
scons: building terminated because of errors.
Using clang++ gives a slightly different error and interesting information
messages I think.
Lastly Christopher has started to move the overlay to use the new python and
distutils eclass starting with sage.
Francois