Uploaded image for project: 'Data Management'
  1. Data Management
  2. DM-10453

Fix bugs in matchPessimisticB

    XMLWordPrintable

Details

    • 2
    • Alert Production S17 - 5
    • Alert Production

    Description

      Bugs have been found in the latest refactor of matchPessimsiticB that effect performance of the matcher.

      failedPatternList persists between runs of MatchPessimisticBTask when running a range of ccds in the same processCcd run.
      failedPatternList is not currently used properly in the matcher loop.
      The pair_id lookup table is not filled with the correct data for half of the matrix.

      Attachments

        Activity

          Created new ticket branch and started search for bugs in the code.

          cmorrison Chris Morrison [X] (Inactive) added a comment - Created new ticket branch and started search for bugs in the code.

          Discovered two bugs in the code.

          The first was that the lastMatchedPattern and failedPatternList were not properly filled or used. The second was more substantial and effected code performance.

          The lookup table the code uses to quickly grab data for each of the candidate spokes (_pair_id_array) was filled incorrectly. Specifically the lower triangular portion of the matrix (which is filled on line 130) was, instead of being filled with the first id of the pair, filled with the second id of the pair. This resulted in half the data being unusable and degraded performance (I ran some CFHT data I had previously and noticed significantly more failures.) The performance has now ben restored with this fix.

          cmorrison Chris Morrison [X] (Inactive) added a comment - Discovered two bugs in the code. The first was that the lastMatchedPattern and failedPatternList were not properly filled or used. The second was more substantial and effected code performance. The lookup table the code uses to quickly grab data for each of the candidate spokes (_pair_id_array) was filled incorrectly. Specifically the lower triangular portion of the matrix (which is filled on line 130) was, instead of being filled with the first id of the pair, filled with the second id of the pair. This resulted in half the data being unusable and degraded performance (I ran some CFHT data I had previously and noticed significantly more failures.) The performance has now ben restored with this fix.
          rowen Russell Owen added a comment -

          Looks good. A few comments in github.

          rowen Russell Owen added a comment - Looks good. A few comments in github.

          Completed code review, Jenkins CI, and pushed to master.

          cmorrison Chris Morrison [X] (Inactive) added a comment - Completed code review, Jenkins CI, and pushed to master.

          People

            cmorrison Chris Morrison [X] (Inactive)
            cmorrison Chris Morrison [X] (Inactive)
            Russell Owen
            Chris Morrison [X] (Inactive), Russell Owen
            Votes:
            0 Vote for this issue
            Watchers:
            2 Start watching this issue

            Dates

              Created:
              Updated:
              Resolved:

              Jenkins

                No builds found.