Details
-
Type:
Bug
-
Status: Done
-
Resolution: Done
-
Fix Version/s: None
-
Component/s: meas_algorithms
-
Labels:None
-
Story Points:1
-
Epic Link:
-
Sprint:AP S19-2
-
Team:Alert Production
Description
In the testFootprintsMeasure function in tests/test_measure.py in meas_algorithms, the test claims to "Check that we can measure the objects in a detectionSet". All of the "asserts" of the test occur within a loop over the sources in measCat (here). HOWEVER, when I run this test, I get 0 detections, so that loop is never entered and no assertions get made (which I noticed when I was not seeing any dots on the sources in ds9 when updating the display code). It seems that something is seriously amiss if we are expecting (it looks like at least 3) detections but getting 0. (I also think I see this as the line
measurement INFO: Measuring 0 sources (0 parents, 0 children)
|
in Jenkins artifact files, e.g. here, so it's not just a local setup issue).
Also, in the testDetection function, nothing is actually "asserted" in the function, so it seems that the "test" is that it does not crash when running through the operations? Would a check on a (minimum?) number of detections be appropriate here?
Attachments
Activity
Hey Lauren, wanna do a quick review?
PR here: https://github.com/lsst/meas_algorithms/pull/146
Jenkins is still running here: https://ci.lsst.codes/blue/organizations/jenkins/stack-os-matrix/detail/stack-os-matrix/29297/pipeline
Minor comments on the PR. Otherwise, if Jenkins is happy, I’m happy. Thanks for the fix.
Responded to review comments; Jenkins is happy; merged & done. Thanks Lauren MacArthur!
For reference, this was broken by 6e52ace (ie, a bit over four years ago!).
The key change was the removal of a call to FootprintSet.makeSources, but there's been a fair bit of other bit-rot since then so just restoring that call doesn't fix everything. I pushed an almost-fixed version which still needs a bit of TLC here; I think it'd be easy enough to finish up, but I'm out of time for today.