Sage 9.1 Release Tour

in progress (2020)

Python 3 transition

SageMath 9.0 was the first version of Sage running on Python 3 by default. Sage 9.1 continues to support Python 2.

See Python3-Switch for more details

Portability improvements, increased use of system packages

The SageMath distribution continues to vendor versions of required software packages ("SPKGs") that work well together.

In order to reduce compilation times and the size of the SageMath installation, a development effort ongoing since the 8.x release series has made it possible to use many system packages provided by the OS distribution instead of building SageMath's own copies.

This so-called "spkg-configure" mechanism runs at the beginning of a build from source, during the ./configure phase.

(See the sage-devel threads "Brainstorming about Sage dependencies from system packages" (May 2017) and "conditionalise installation of many spkg's?" (Nov 2017) for its origins and Trac #24919 for its initial implementation.)

Sage 9.1 is adding many packages to this mechanism, including openblas, gsl, r, boost, libatomic, cddlib, tachyon, nauty, sqlite, planarity, fplll, brial, flintqs, ppl, libbraiding, cbc, gfan, and python3. As to the latter, SageMath will now make use of a suitable installation of Python 3.7.x in your system by setting up a venv (Python 3 virtual environment).

New in Sage 9.1 is also a database of system packages equivalent to our SPKGs. At the end of a ./configure run, you will see messages like the following:

  configure: notice: the following SPKGs did not find equivalent system packages: arb boost boost_cropped bzip2 ... yasm zeromq zlib
  checking for the package system in use... debian
  configure: hint: installing the following system packages is recommended and may avoid building some of the above SPKGs from source:
  configure:   $ sudo apt-get install libflint-arb-dev ... yasm libzmq3-dev libz-dev
  configure: After installation, re-run configure using:
  configure:   $ ./config.status --recheck && ./config.status

For developers

For developers who wish to help improve the portability of SageMath, there is a new power tool: A tox configuration that automatically builds and tests SageMath within Docker containers running various Linux distributions (ubuntu-trusty through -focal, debian-jessie through -sid, linuxmint-17 through -19.3, fedora-26 through -32, centos-7 and -8, archlinux, slackware-14.2), each in several configurations regarding what system packages are installed. Thus, it is no longer necessary for developers to have access to a machine running fedora-29, say, to verify whether the Sage distribution works there; instead, you just type:

  tox -e docker-fedora-29-standard -- build ptest

The Dockerfiles are generated automatically by using the same database of system packages that provides information to users. See the new section on "portability testing" in the Developer's Guide for details.

An entry point for developers who wish to improve the testing infrastructure is the Meta-Ticket #29060: Add Dockerfiles and CI scripts for integration testing. See also the broader Meta-Meta-Ticket #29133.

Add more here

(add more here)

Easier installation of optional linear and mixed integer linear optimization backends

It is no longer necessary to recompile sagelib if you wish to use one of the state-of-the-art LP/MIP solvers COIN-OR CBC, CPLEX, or Gurobi, instead of the default (GLPK). The simplified new installation procedure is explained in the Thematic Tutorial on Linear Programming.

(If you cannot update to 9.1 just yet, you can retroactively get the same feature in your installation of Sage too by pip-installing one of the packages sage-numerical-backends-cplex, sage-numerical-backends-coin, sage-numerical-backends-gurobi.)

Polyhedral Geometry

There is now a catalog for common polyhedral cones, e.g.

sage: cones.nonnegative_orthant(5)
5-d cone in 5-d lattice N

New features for polyhedra:

sage: P = polytopes.cube(intervals='zero_one') # obtain others than the standard cube
sage: P = matrix([[0,1,0],[0,1,1],[1,0,0]])*P  # linear transformations
sage: it = P.face_generator()                  # a (fast and efficient) face generator
sage: next(it)
A 3-dimensional face of a Polyhedron in ZZ^3 defined as the convex hull of 8 vertices
sage: next(it)
A -1-dimensional face of a Polyhedron in ZZ^3
sage: f = next(it)
sage: f.normal_cone()                          # normal cone for faces
A 1-dimensional polyhedron in ZZ^3 defined as the convex hull of 1 vertex and 1 ray
sage: P.an_affine_basis()                      # an_affine_basis and a_maximal_chain
[A vertex at (0, 0, 0),
 A vertex at (1, 1, 0),
 A vertex at (0, 0, 1),
 A vertex at (0, 1, 0)]
sage: P = polytopes.hypercube(4)
sage: P.flag_f_vector(0,3)                     # flag_f_vector is exposed
64

Regarding the optional package normaliz there are some news as well:

sage: P = polytopes.cube(intervals=[[0,1],[0,2],[0,3]], backend='normaliz')
sage: save(P, '/tmp/this_takes_very_long_so_we_save_it')   # saving works now
sage: sage: P.h_star_vector()                              # compute the h_star_vector with normaliz
[1, 20, 15]

There are also some bug fixes and other improvements. For more details see the release notes for optimization and polyhedral geometry softwares interactions in Sage.

Add more here

(add more here)

Availability of Sage 9.1 in distributions

(TBD)

More details

- Release announcements: See https://groups.google.com/forum/#!forum/sage-release

- Trac tickets with milestone 9.1