Please allow sagemath into Lenny
Today the sagemath package and necessary improvements to 4 of its
dependencies were uploaded. I was going to do the uploads on Friday, but
was unable to get in touch with my usual sponsor, Christine Spang. I
asked Karl Ramm to sponsor the uploads Saturday night, but he didn't get a
chance to upload them until today. So, I missed the freeze.
There are 8 uploads involved.
Two are small bugfixes that I think would be eligible for a freeze
exception normally (I've not reported the bugs, since I found them myself):
- guava_3.6-2.dsc (a trivial bugfix)
- debdiff at <http://web.mit.edu/tabbott/Public/guava-3.6-2.debdiff>
- sympow_1.019-3.dsc (fixing support for generating datafiles, and
including by default a standard set that upstream recommends. The
package is unusable without these)
- truncated debdiff at
<http://web.mit.edu/tabbott/Public/sympow_1.019-3.debdiff>,
(I removed the datafiles/datafile generation code from the debdiff
since they're long and obscure the actual changes. Full debdiff at
http://web.mit.edu/tabbott/Public/sympow_1.019-3.debdiff-full).
One is a small upstream update that is required for sagemath; it adds two
new API functions and makes some minor changes to internal functions (I
tried for a while to use the nullspaceLong function rather than the (new)
nullspaceMP function, but it is not really possible). I chose to take the
upstream update rather than patching it in myself to avoid getting into
library versioning trouble.
- iml-1.0.3-1.dsc
Four are packages that have been sitting in NEW since July 16. If the
automatic exception for packages in unstable as of the freeze doesn't
apply to packages that were in NEW then, I'd like an exception for them
(assuming of course that they are accepted).
- polybori_0.5~rc1-1.dsc
- libzn-poly_0.8-1.dsc
- singular_3-0-4-3.dfsg-1.dsc
- linbox_1.1.6~rc0-2.dsc (linbox_1.1.6~rc0-1.dsc was in NEW since July 16.
This version fixes a missing build-dependency and applies
workarounds recommended by upstream for a problem with their
commentator system that broke substantial core functionality and
a problem with their default characteristic polynomial algorithm.
The update from -1 to -2 is just bugfixes would normally be eligible for
a freeze exception).
and then there is the sagemath package itself, which is just plain late.
- sagemath_3.0.5dfsg-1.dsc (uploaded to NEW today)
I'm fairly pleased with the quality of the sagemath package; I fixed about
20 significant bugs in the packaging during the week before Friday. Sage
has over 10000 doctests (covering more than half of the functions in
Sage), and the current package works correctly on all of them except for
the external interfaces to qd and polybori (because Debian's versions of
those packages are newer than those the Sage interfaces are designed for).
Upstream is planning to completely drop the qd interface in the near
future because nobody uses it
(http://groups.google.com/group/sage-devel/browse_thread/thread/4bc3ac8ed1a53fb7).
The polybori interface I would like to fix in a bugfix upload later, once
upstream has a patch for it, but it is one of the less important
interfaces and there's always the workaround of directly using polybori's
built-in python module.
The other significant known bug affecting the sagemath package is that
some databases of elliptic curves (etc.) aren't accessible because of
#472392 (which should probably be release-critical anyway). The sagemath
package does have around 50 Lintian warnings, but they are all
non-functional problems (like README files incorrectly marked executable)
that I think are best fixed upstream (and upstream says they will fix).
Sage has gotten substantial press recently because it is the only free
mathematics software program that is seriously competing with Mathematica
(see <http://wiki.sagemath.org/SAGE_in_the_News> for a number of articles
related to it, including coverage on slashdot, reddit, etc.). Sage is a
huge pain to install from source (it takes about 8 hours to build, because
it comes with 70 dependencies, 25 of which weren't available in Debian
when I started working on this 7 months ago), and I think the Debian
science/math community would greatly benefit from it being in Lenny.
None of these packages have reverse dependencies other than sagemath, and
the popcon numbers for all of them are at most 10 (which I suspect will
change only when sagemath gets into Debian, since many of these libraries
are primarily used via Sage even by their authors). Thus I think
accepting these uploads a couple days after the freeze is unlikely to
cause problems for the release.
Please consider making an exception for sagemath to enter Lenny.
-Tim Abbott
Reply to: