Details
-
Type:
Bug
-
Status: Done
-
Resolution: Done
-
Fix Version/s: None
-
Component/s: meas_algorithms
-
Labels:None
-
Story Points:1
-
Epic Link:
-
Team:Alert Production
Description
Given the recent change to how interpolated is used in the stack, it appears that BaseSourceSelectorConfig should not include interpolated in its list of default badFlags. We may want to add any new "why interplated" flags instead.
Attachments
Issue Links
- relates to
-
DM-14262 Cleanup fallout from removing interpolated from badFlags list
- To Do
Activity
Jenkins failed in Jointcal. 
Looks like that's because the Jointcal tests hard-code exactly how many stars should be selected. Of course, one might expect that those numbers would be different given this change, so I think it's ok to just change the test values... but I shall defer to my distinguished reviewer before doing that. John Parejko, any thoughts?
Yes, those tests are there specifically to catch changes in the source selectors. I think you're going to want to run jointcal yourself and look at the output to make sure the number of sources is still reasonable in the tests, otherwise we may have to reprocess the test data to bring it in line with the new definition of interpolated.
The fact that the number of sources has changed will likely also affect the chi2, ndof and other values, which is why I recommend running jointcal yourself to catch all the changes.
Unfortunately, this is no longer a ticket I can handle in between agenda and milestone munging. Back on the backlog!
I realised that, rather than updating all the test results to reflect the fact that we're now accepting interpolated pixels, we can simply update the tests to continue to reject them. I think this is appropriate, but perhaps it should be accompanied by another ticket to refresh the test data and then revert this change to the tests.
Anyway:
- PR on meas_algorithms: https://github.com/lsst/meas_algorithms/pull/103
- PR on Jointcal: https://github.com/lsst/jointcal/pull/68
- Success on Jenkins: https://ci.lsst.codes/blue/organizations/jenkins/stack-os-matrix/detail/stack-os-matrix/27473/
Whatcha reckon, John Parejko?
Sorry for the delay in reviewing. These look fine: please file a ticket on refreshing the testdata (minor priority, probably).
I did see the interpolated flag show up in pipe.tasks.photoCal.py and meas.algorithms.starSelector: it should definitely be removed from the latter, but I'm not sure about the former.
Pinging this ticket: it's reviewed and can probably be merged after it's rebased to master.
Sorry for slowness; now merged.
As promised...
PR here: https://github.com/lsst/meas_algorithms/pull/103
Jenkins here: https://ci.lsst.codes/blue/organizations/jenkins/stack-os-matrix/detail/stack-os-matrix/27469/pipeline/ (not finished at time of posting, but I have faith...)