no checking on getitem for Ambient space of a root system

[VivianePons]

When you create an ambient space of a root system, there is a getitem method. But no verification is made on the given value which implies bugs on other uses of the object.

sage: R = RootSystem('A2')
sage: A2 = R.ambient_space()
sage: A2[0]
(0, 0, 0)
sage: list(A2[0])
[(-1, 1)]
sage: A2.zero()
(0, 0, 0)
sage: list(A2.zero())
[]

The getitem should only get values between 1 and 3 and raise exceptions when not. Otherwise, it allows for creating elements which don't make any sense and can bring trouble :

sage: C = CombinatorialFreeModule(QQ,A2)
sage: C.an_element()
B[(0, 0, 0)] + 2*B[(1, 0, 0)] + B[(2, 2, 3)] + 3*B[(0, 1, 0)]
sage: elt = C(A2([0,0,0])) + 2 * C(A2([1,0,0])) + C(A2([2,2,3])) + 3 * C(A2([0,1,0]))
sage: elt
B[(0, 0, 0)] + 2*B[(1, 0, 0)] + B[(2, 2, 3)] + 3*B[(0, 1, 0)]
sage: elt == C.an_element()
False

Indeed, the getitem is used by python to iterate on the parent and the iteration is used by CombinatorialFreeModule? to create elements.

[nthiery]

Following the discussion on

​http://groups.google.com/group/sage-combinat-devel/browse_thread/thread/ab1722fb97c0fbbb

this is good to go (assuming the buildbot goes green which I expect)!

Thanks,

Nicolas

[jdemeyer]

@VivianePons?: create a .hgrc file as explained in ​http://sagemath.org/doc/developer/walk_through.html#submitting-a-change and resubmit the patch.

[nthiery]

Replying to jdemeyer:

@VivianePons?: create a .hgrc file as explained in ​http://sagemath.org/doc/developer/walk_through.html#submitting-a-change and resubmit the patch.

Sorry, I should have caught this. The main point is that the patch should be exported before being posted on trac. See also: ​http://wiki.sagemath.org/combinat/MercurialStepByStep#ExportingPatchesforusewithtrac.

[VivianePons]

the patch