phi(I) for phi a ring morphism and I an ideal should work (IMHO); it used to and now it doesn't because of newish arithmetic architecture stuff
Who rewrote the ring morphism code so that if phi is a
ring morphism, then phi(I) no longer works, for an ideal I?
Oh, David Roed in changeset 6772 (for me) from a few
months ago did this:
"Cython'ed sage/rings/morphism.py, actually added wrapper_parent (even though I claimed to in the previous commit)."
I think that feature, i.e., that phi(I) works, was very nice
and is standard notation in mathematics, and I want
it back. Then the codepath that leads to the above weird
bug wouldn't exist.
I think the way to fix this is:
(1) Rethink the assumption you're forcing on morphisms that they
can only apply to elements in the domain. This overloading of
calling a morphism on (sub)objects is very standard in mathematics.
(2) Change the architecture of __call__ as a result of (1).
 William
Change History (2)
Resolution: 
→ worksforme

Status: 
new →
closed

Milestone: 
sage3.4.1 →
sage3.3

This was previously fixed. See sage.categories.map.Map.call