Ring morphisms should not override __richcmp__

[saraedum]

But only override _richcmp_ instead.

Overwriting __richcmp__ used to be necessary to override __hash__. This is not the case anymore it seems.

CC: @roed314

Component: refactoring

Keywords: sd87

Author: Julian Rüth

Branch/Commit: ffc7371

Reviewer: Aly Deines, Jeroen Demeyer, David Roe

Issue created by migration from https://trac.sagemath.org/ticket/23485

[saraedum]

Branch: u/saraedum/ring_morphisms_should_not_override___richcmp__

[saraedum]

Changed keywords from none to sd87

[saraedum]

Commit: 2ee2abf

[saraedum]

New commits:

2ee2abf

Do not override `__richcmp__` in Ring morphisms

[jdemeyer]

comment:5

These blocks of code should go away:

        if not isinstance(other, RingHomomorphism_cover):
            return (op == Py_NE)

I disagree with them in the first place and they are just pointless now.

[sagetrac-git]

Changed commit from 2ee2abf to b6ef396

[sagetrac-git]

Branch pushed to git repo; I updated commit sha1. New commits:

b6ef396

Remove obsolete type checks

[adeines]

comment:8

Passes all doctests now too.

[adeines]

Reviewer: Aly Deines

[jdemeyer]

comment:9

Sorry, I was wrong. At least some of the type checks are still needed, because equal parents does not imply equal types here. For example:

sage: R.<x,y> = PolynomialRing(QQ, 2)
sage: S.<a,b> = R.quo(x^2 + y^2)
sage: phi = S.cover()
sage: alpha = R.hom(R, (0,0))
sage: psi = phi.pre_compose(alpha)
sage: parent(psi) is parent(phi)
True
sage: type(psi) is type(phi)
False

[jdemeyer]

Changed reviewer from Aly Deines to Aly Deines, Jeroen Demeyer

[sagetrac-git]

Changed commit from b6ef396 to ffc7371

[sagetrac-git]

Branch pushed to git repo; I updated commit sha1. New commits:

ffc7371

Revert "Remove obsolete type checks"

[roed314]

comment:12

Looks good to me.

[roed314]

Changed reviewer from Aly Deines, Jeroen Demeyer to Aly Deines, Jeroen Demeyer, David Roe

[vbraun]

Changed branch from u/saraedum/ring_morphisms_should_not_override___richcmp__ to ffc7371