Uploaded image for project: 'Data Management'
  1. Data Management
  2. DM-8046

whatToFit should use an enumlist/bitfield, not string parsing

    Details

      Description

      Jointcal's AstromFit::AssignIndices (and subsequent methods) use a string, whatToFit that contains various things defining what parameters to fit (e.g. "Distortions", "Positions", "DistortionPositions"). This could probably be done better with an enum list or bit field, so we don't have to parse the string.

        Attachments

          Issue Links

            Activity

            Hide
            Parejkoj John Parejko added a comment - - edited

            As part of this work, I should also consolidate the astrometry/photometry code to both use Model and Values, so that I can consolidate some high level code (e.g. _fit_photometry() and _fit_astrometry) to reduce duplication.

            Show
            Parejkoj John Parejko added a comment - - edited As part of this work, I should also consolidate the astrometry/photometry code to both use Model and Values , so that I can consolidate some high level code (e.g. _fit_photometry() and _fit_astrometry ) to reduce duplication.
            Hide
            Parejkoj John Parejko added a comment -

            Krzysztof Findeisen made this comment when I introduced _logChi2AndValidate():

            So, this is a pretty blatant violation of the "do one thing" rule. I can understand moving the logger and finiteness tests in here (though logging is not normally considered part of a function/program's spec, it just happens), but why can't you have a separate call for validating the model?

            When I consolidate the Model/Values naming convention, I can also try to clean up the above method, since there will be fewer total calls to manage.

            Show
            Parejkoj John Parejko added a comment - Krzysztof Findeisen made this comment when I introduced _logChi2AndValidate() : So, this is a pretty blatant violation of the "do one thing" rule. I can understand moving the logger and finiteness tests in here (though logging is not normally considered part of a function/program's spec, it just happens), but why can't you have a separate call for validating the model? When I consolidate the Model/Values naming convention, I can also try to clean up the above method, since there will be fewer total calls to manage.

              People

              • Assignee:
                Unassigned
                Reporter:
                Parejkoj John Parejko
                Watchers:
                John Parejko, Simon Krughoff
              • Votes:
                0 Vote for this issue
                Watchers:
                2 Start watching this issue

                Dates

                • Created:
                  Updated:

                  Summary Panel