Uploaded image for project: 'Data Management'
  1. Data Management
  2. DM-21000

obs_base has code that depends on pipe_base

    Details

    • Team:
      Data Release Production

      Description

      obs_base has code, at least https://github.com/lsst/obs_base/blob/master/python/lsst/obs/base/gen3/ingest.py, which depends on pipe_base, which is not captured by the eups table.

       

      Unfortunately, pipe_base depends on obs_test (via setupOptional) which depends on obs_base, so we get a cycle.

       

       

        Attachments

          Issue Links

            Activity

            Hide
            krughoff Simon Krughoff added a comment - - edited

            I also worry that afwdata may not be the right place due to its size and the way that it has grown organically in the past. The standard solution does seem to be testdata_obs_test and I don't have a particular objection to that in this case.

            I am strongly opposed to putting test data in obs_test_data. I don't want to conflate test data with calibration data.

            Show
            krughoff Simon Krughoff added a comment - - edited I also worry that afwdata may not be the right place due to its size and the way that it has grown organically in the past. The standard solution does seem to be testdata_obs_test and I don't have a particular objection to that in this case. I am strongly opposed to putting test data in obs_test_data . I don't want to conflate test data with calibration data.
            Hide
            tjenness Tim Jenness added a comment -

            We've had another look at this and we can't simply use the data from obs_test because we need to read the butler repository and the mapper in that repository is the obs test mapper. I think the only solution for pipe_base is to have a very simple gen2 repo in the git repository with only one file and a very simple mapper.

            Show
            tjenness Tim Jenness added a comment - We've had another look at this and we can't simply use the data from obs_test because we need to read the butler repository and the mapper in that repository is the obs test mapper. I think the only solution for pipe_base is to have a very simple gen2 repo in the git repository with only one file and a very simple mapper.
            Hide
            jbosch Jim Bosch added a comment -

            IMO the right solution here is to move pipe_base tests that depend on obs_test to a new test-script-only package.  That's not a pattern we have a precedent for, but it's one that would help out in many more places; it's arguably what we should always do for tests of packages that are intended to provide abstract interfaces with no concrete implementations.

            Show
            jbosch Jim Bosch added a comment - IMO the right solution here is to move pipe_base tests that depend on obs_test to a new test-script-only package.  That's not a pattern we have a precedent for, but it's one that would help out in many more places; it's arguably what we should always do for tests of packages that are intended to provide abstract interfaces with no concrete implementations.
            Hide
            bvan Brian Van Klaveren added a comment -

            yeah I was about to start creating a `pipe_test` repo. I do believe the `obs_test` repo is actually, in a way, written according to that pattern, though it makes sense in any case to use this pattern, especially when the tests are more integrated with multiple packages, as this case is.

            Show
            bvan Brian Van Klaveren added a comment - yeah I was about to start creating a `pipe_test` repo. I do believe the `obs_test` repo is actually, in a way, written according to that pattern, though it makes sense in any case to use this pattern, especially when the tests are more integrated with multiple packages, as this case is.
            Hide
            jbosch Jim Bosch added a comment -

            I think this problem has been solved on DM-21351, which just moved the offending tests to obs_test itself, so I'm closing this ticket as a duplicate.  If anyone wants to instead move them to a new package, we can reopen this ticket for that.

             

            Show
            jbosch Jim Bosch added a comment - I think this problem has been solved on DM-21351 , which just moved the offending tests to obs_test itself, so I'm closing this ticket as a duplicate.  If anyone wants to instead move them to a new package, we can reopen this ticket for that.  

              People

              • Assignee:
                Unassigned
                Reporter:
                bvan Brian Van Klaveren
                Watchers:
                Brian Van Klaveren, Jim Bosch, John Parejko, Simon Krughoff, Tim Jenness
              • Votes:
                1 Vote for this issue
                Watchers:
                5 Start watching this issue

                Dates

                • Created:
                  Updated:
                  Resolved: