Uploaded image for project: 'Data Management'
  1. Data Management
  2. DM-13836

Specify atol in calls to numpy.testing.assert_allclose

    Details

    • Type: Improvement
    • Status: Done
    • Resolution: Done
    • Fix Version/s: None
    • Component/s: astshim
    • Templates:
    • Story Points:
      0.25
    • Sprint:
      AP S18-6, AP F18-1, AP F18-2
    • Team:
      Alert Production

      Description

      For numpy.testing.assert_allclose the default value of the atol argument is 0. This is a bad choice for values that are supposed to be 0, as even irrelevant machine differences can cause a test failure.

      This caused a failure in astshim one one machine (but not any of our standard machines) as reported here: https://community.lsst.org/t/install-package-without-tests/2746/5

      I suggest specifying a reasonable non-zero value of atol in all our usage of assert_allclose in order to make our tests more robust.

        Attachments

          Activity

            People

            • Assignee:
              rowen Russell Owen
              Reporter:
              rowen Russell Owen
              Watchers:
              Kian-Tat Lim, Russell Owen
            • Votes:
              0 Vote for this issue
              Watchers:
              2 Start watching this issue

              Dates

              • Created:
                Updated:
                Resolved:

                Summary Panel