Opened 10 years ago

Closed 10 years ago

Last modified 10 years ago

#11277 closed enhancement (fixed)

Temporarily disable failing SVD doctest

Reported by: jdemeyer Owned by: mvngu
Priority: blocker Milestone: sage-4.7
Component: doctest coverage Keywords:
Cc: Merged in: sage-4.7.rc1
Authors: Jeroen Demeyer Reviewers: Rob Beezer
Report Upstream: N/A Work issues:
Branch: Commit:
Dependencies: Stopgaps:

Status badges

Description

On one particular configuration (Mac, gcc 4.2.1), there is a doctest failure in sage-4.7.alpha5 (See http://groups.google.com/group/sage-release/msg/d29ce3e4f9da5cda):

sage -t  -long -force_lib "devel/sage/sage/matrix/matrix_double_dense.pyx"
**********************************************************************
File "/Applications/sage-4.7.alpha5/devel/sage/sage/matrix/matrix_double_dense.pyx",
line 1390:
    sage: V.is_unitary()
Expected:
    True
Got:
    False
********************************************************************** 

This issue is #11248, but I propose to temporarily disable this doctest to be able to ship sage-4.7 without failing doctests (otherwise, people will complain).

Attachments (1)

11277.patch (2.5 KB) - added by jdemeyer 10 years ago.

Download all attachments as: .zip

Change History (7)

Changed 10 years ago by jdemeyer

comment:1 Changed 10 years ago by jdemeyer

  • Authors set to Jeroen Demeyer
  • Status changed from new to needs_review

comment:2 Changed 10 years ago by rbeezer

  • Status changed from needs_review to positive_review

Applies, builds, passes tests. However, I do not have a configuration where it was failing.

I did test that the test was not executed - by totally breaking it with a non-existent command, so the comment in place ("not tested") is doing its job.

Two comments:

1) This file needs lots more documention help (it is not in the reference manual yet).

2) Once this test is restored, we should really test the SVD, that seems to be where the true bug is.

comment:3 Changed 10 years ago by jdemeyer

  • Merged in set to sage-4.7.rc1
  • Resolution set to fixed
  • Reviewers set to Rob Beezer
  • Status changed from positive_review to closed

comment:4 follow-ups: Changed 10 years ago by drkirkby

If the platform is getting an error, why should the test be disabled? That seems to defeat the object of a test to me.

People might complain about a test failure, but at least we can make them aware of the problem. Instead this is covered up.

Many methods of testing software have "expected failures" where a test is known to fail. We could really do with something like that in Sage. But I really don't like the idea of disabling a test on every platform, just so all the tests pass on one platform.

Dave

comment:5 in reply to: ↑ 4 Changed 10 years ago by jdemeyer

Replying to drkirkby:

People might complain about a test failure, but at least we can make them aware of the problem. Instead this is covered up.

Given that it has a Trac ticket, it is not really "covered up". After disabling the test as in this ticket, the SVD bug is no different from most other Sage bugs: known in Trac but doesn't cause doctest failures.

If we want to use your logic, every known Sage bug should have a doctest demonstrating that bug.

comment:6 in reply to: ↑ 4 Changed 10 years ago by rbeezer

Replying to drkirkby:

If the platform is getting an error, why should the test be disabled? That seems to defeat the object of a test to me.

Hi Dave,

The test was meant to document is_unitary(). It seems to work.

The SVD would appear to have been broken for a long time. The bad output from the SVD was caught by the unitary test.

If anything, we would add a test to the SVD routine. But practice has been to add those tests once they are fixed. There is extensive documentation of the bug at #11248.

Thanks, Rob

Note: See TracTickets for help on using tickets.