This appears to be a fairly good start toward a system of installation
that I have been interested in for several years. using the packages
already available in Debian or other linux would go a LONG way towards
mainstreaming Sage. I have the proper environment to experiment in & am
willing to do so as a tester. Best wishes!!!
On 02/04/2015 08:30 AM, Julien Puydt wrote:
Hi,
here is a more precise explanation of my experiments.
First, let's see how I do a testing run :
!!! BEWARE the following commands clean part of your home dir then
install cruft in your home dir!!!
Would this work better by creating a USER: sage account on your system,
thereby defining the permissions of the installation and not causing any
issues with your current /home/youruser account.
(1) rm -rf sage-exp .local/lib/python2.7/site-packages/sage*
(cleaning the previous attempt)
(2) cp -R sage.git sage-exp
(getting the upstream source)
(3) cd sage-exp
(well...)
(4) patch -p1 < ~/unbreak.patch
(well...)
(5) python ~/pruner.py
(see below about that pruner script)
Can you provide more insight into the pruner.py script; i.e. how to
obtain it or create it. (I have no python skills, but am pretty well
versed in Debian.)
Oh I haven't examined the attachment yet, but will be following up with
that.
I find this all very exciting.
(6) export MAKE="make -k -j 3"
(not sure the -k is needed anymore)
(7) export SAGE_ATLAS_LIB="/usr/lib"
(not sure it's needed at all)
(8) make ptestlong
(will build sage, install parts of it in ~/.local, compile the
documentation, then run the tests)
The pruner python script is obtained by running a configure script ;
its goal is to find out what is on the system and make sage use that
instead of its own version. You're not supposed to try all
combinations of have/haven't : it doesn't do anything smart with deps.
The current version prints the following for me:
libgap will not come from debian!
cddlib will not come from debian!
singular will not come from debian!
jmol will not come from debian!
95 system packages will be used
So if it complains about something else not coming from debian, you
should have a look at why it thinks you don't have it, and get it.
Some packages need special tricks :
- sagenb comes from the debian-science git repository ;
- lcalc is available in debian, but you'll need the +dfsg-4 I pushed
in the debian-science repository ;
- rubiks is available since this morning in the debian-science
repository ; I'll have to review it and ask for sponsorship later ;
- singular (3-1-6) from the debian-science needs to be installed, even
if sage compiles its own (it's probably wrong, but I don't really care
since I'm waiting for a good package) ;
- maxima is quite complex to get going. The usual debian packages
should be installed first to make configure happy, but then some more
needs to be done : start from a debian package source, apply sage's
build-fasl.patch patch, add --enable-ecl to the debian/rules configure
line, "make -f debian/rules binary" (will fail unless root : not a
problem!), then copy (as root) src/binary-ecl/maxima.fas as
/usr/lib/ecl-13.5.1/maxima.fas. To see if that worked, launch ecl the
following should tell you it managed to load things :
(require 'maxima)
(in-package maxima)
(set-pathnames)
#$load(simplify_sum)$
There are still quite a few problems with the maxima+ECL combination :
a few tests fail because the output looks like what sage expects on 32
bits while I'm on 64 bits, other because of some crash... I'm still
working on it.
You'll find attached everything needed to reproduce the above.
Snark on #debian-science
--
John Foster
JW Foster & Associates
--
You received this message because you are subscribed to the Google Groups
"sage-devel" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to sage-devel+unsubscr...@googlegroups.com.
To post to this group, send email to sage-devel@googlegroups.com.
Visit this group at http://groups.google.com/group/sage-devel.
For more options, visit https://groups.google.com/d/optout.
