(→Missing required components: rm lidia: only optional dependency of LinBox, unacceptable license (NC), not shipped by SAGE itself (probably because of the license)) |
(→Missing required components: Add note about SAGE's Symmetrica packaging.) |
||
Line 71: | Line 71: | ||
* [http://www.singular.uni-kl.de/ Singular] | * [http://www.singular.uni-kl.de/ Singular] | ||
* [http://www.maths.bris.ac.uk/%7Emamjw/ Sympow] | * [http://www.maths.bris.ac.uk/%7Emamjw/ Sympow] | ||
* [http://www.mathe2.uni-bayreuth.de/axel/symneu_engl.html Symmetrica] This will be a challenge because upstream didn't really design it to be distributed at all. | * [http://www.mathe2.uni-bayreuth.de/axel/symneu_engl.html Symmetrica] This will be a challenge because upstream didn't really design it to be distributed at all. <ref>SAGE simply ships a libsymmetrica.a with all the source files built and collected into and puts the 2 header files (def.h and macro.h) into include/symmetrica. We should do the same to be compatible, except possibly shipping the library as shared instead of static, but then we'll have to deal with soname versioning.</ref> | ||
* [http://code.google.com/p/sympy/ Sympy] [https://bugzilla.redhat.com/show_bug.cgi?id=466660 Review Request Completed, in Fedora] | * [http://code.google.com/p/sympy/ Sympy] [https://bugzilla.redhat.com/show_bug.cgi?id=466660 Review Request Completed, in Fedora] | ||
* [http://www.scipy.org/Weave Weave] Already in scipy. | * [http://www.scipy.org/Weave Weave] Already in scipy. |
Revision as of 17:51, 7 December 2008
This is to help document the process to get SAGE (http://sagemath.org/), the open-source computer algebra/mathematics framework into Fedora.
People interested helping in packaging/reviewing:
- Rex Dieter
- Alex Lancaster
- Manuel Wolfshant
- Kevin Kofler
- Neal Becker
- Conrad Meyer
- ...add your name here...
Cross-distribution efforts:
- Debian: http://wiki.sagemath.org/DebianSAGE
- PcLinuxOS: http://www.mypclinuxos.com/forum/index.php?topic=1509.msg13532#msg13532
On fedora-devel KevinKofler points out that:
One problem will be that they're bundling many third-party components which should be packaged separately:
http://sage.math.washington.edu/sage/doc/html/inst/intro.html
So the first step is to track down which of these dependencies are in Fedora already, whether they need any patches to work with SAGE, whether they are build-time (BuildRequires) dependencies, run-time (Requires) dependencies or both, whether they're required or optional and package those which are not in Fedora yet.
He suggests the following order:
- package required build-time dependencies
- package required run-time dependencies
- package as many optional build-time dependencies as possible
- package SAGE itself
- package optional run-time dependencies (and decide on a case by case basis whether it makes sense to add them as actual Requires: dependencies to the package or not)
Required components that SAGE ships with: http://sagemath.org/packages/standard/
Optional components that SAGE ships with: http://sagemath.org/packages/optional/
(a manual list with descriptions which may vary somewhat with respect to the above is at http://sagemath.org/links-components.html) (See also: http://wiki.sagemath.org/days6/sprint/debian)
Missing required components
(If you are working on packaging a component or have filed a review request bug, please note it or link to something here.)
- Cddlib Review Request
- CVXOPT Review Request
- Flint Review Request
- FpLLL Review Request Completed, in Fedora
- G95 Not needed, we can use gcc-gfortran instead. [1]
- GAP This is large (they claim you need 500M of disk)
- Genus2reduction Wants PARI (Pari is already in Fedora, whoops).
- Gfan Review Request
- Givaro Review Request
- GMP-ECM Review Request completed, in Fedora
- JsMath
- IML Review Request Completed, in Fedora
- Lcalc Review Request
- Linbox Depends on Givaro.
- M4RI Review Request complete, in Fedora
- MPFI Review Request Completed, in Fedora
- EClib
- NetworkX
- PALP
- PolyBoRi
- PyCrypto
- Qd Review Request
- Singular
- Sympow
- Symmetrica This will be a challenge because upstream didn't really design it to be distributed at all. [2]
- Sympy Review Request Completed, in Fedora
- Weave Already in scipy.
- ZODB
- ↑ From upstream's readme.txt: "If you're using Fortran on a platform without g95 binaries included with Sage, e.g., Itanium, you must use a system-wide gfortran.". So they support gfortran and even system-wide installations of it.
- ↑ SAGE simply ships a libsymmetrica.a with all the source files built and collected into and puts the 2 header files (def.h and macro.h) into include/symmetrica. We should do the same to be compatible, except possibly shipping the library as shared instead of static, but then we'll have to deal with soname versioning.