inconsistent error messages for inverting singular matrices

[sagetrac-dmharvey]

n = 2:

sage: Matrix([[1, 1], [1, 1]])^(-1)
...
ZeroDivisionError: 

apparently all other n:

sage: Matrix([[1, 1, 1], [1, 1, 1], [1, 1, 1]])^(-1)
...
ZeroDivisionError: input matrix must be nonsingular

Apply trac_10028_error_messages.patch

Component: linear algebra

Author: André Apitzsch

Reviewer: Rob Beezer

Merged: sage-4.6.2.alpha4

Issue created by migration from https://trac.sagemath.org/ticket/10028

[rbeezer]

comment:2

I just noticed these weak error messages the other day - it'd be great to have all the cases of non-invertible matrices give a common, and informative, message. Two suggestions:

(a) At the sage prompt run search_src("ZeroDivisionError") and I think you will find two or three more places in the matrix code that could also be improved.

(b) For Python 3 raising an error will require syntax like:

raise ZeroDivisionError("input matrix must be nonsingular") 

Might as well do it that way now.

Rob

[a-andre]

Attachment: trac_10028_error_messages.patch.gz

[a-andre]

comment:3

The new patch contains a lot of small changes (hopefully not too much), most of them are related to python3 syntax.

[rbeezer]

Reviewer: Rob Beezer

[rbeezer]

comment:4

Looks real good. Thanks for straightening this out. Passes all tests on 4.6.2.alpha2. Positive review.

[rbeezer]

Author: Andre Apitzsch

[jdemeyer]

Merged: sage-4.6.2.alpha4

[jdemeyer]

Changed author from Andre Apitzsch to André Apitzsch