Uploaded image for project: 'Data Management'
  1. Data Management
  2. DM-829

Algorithm API without (or with optional) Result objects

    XMLWordPrintable

Details

    Description

      In this design prototype, I'll see how much simpler things could be made by making the main algorithm interface one that sets record values directly, instead of going through an intermediate Result object. Ideally the Result objects would still be an option, but they may not be standardized or reusable.

      Attachments

        Activity

          jbosch Jim Bosch added a comment -

          The new interface prototype is complete: I've added new API base classes and some helper classes, and converted PsfFlux and SdssShape to the new interface. Many of the new classes aren't adequately documented, but I wasn't planning to merge this ticket when it's done anyway: that's what DM-1130 is for (I do expect we'll start that branch from this one, however).

          Perry and I already did a live review of most of the new code, but I'd like him to take a look at the wrappers for SdssShape, which are new. They're somewhat more complex that the wrappers for PsfFlux, because they provide an interface for calling the algorithm outside the measurement framework (which isn't needed for PsfFlux). Perry, once you feel you understand the SdssShape wrappers, and noted any obvious problems you find, please move this to Review Complete. I think any deep modifications to this interface that you want to try belong on DM-828 or its subtasks (but you probably want to start those branches from this one).

          Once DM-828 is done, we should put together a Confluence page describing the design, and email lsst-data for a broader design review (I hope we can do that by email, but we'll try to schedule a live session if we have to).

          jbosch Jim Bosch added a comment - The new interface prototype is complete: I've added new API base classes and some helper classes, and converted PsfFlux and SdssShape to the new interface. Many of the new classes aren't adequately documented, but I wasn't planning to merge this ticket when it's done anyway: that's what DM-1130 is for (I do expect we'll start that branch from this one, however). Perry and I already did a live review of most of the new code, but I'd like him to take a look at the wrappers for SdssShape, which are new. They're somewhat more complex that the wrappers for PsfFlux, because they provide an interface for calling the algorithm outside the measurement framework (which isn't needed for PsfFlux). Perry, once you feel you understand the SdssShape wrappers, and noted any obvious problems you find, please move this to Review Complete. I think any deep modifications to this interface that you want to try belong on DM-828 or its subtasks (but you probably want to start those branches from this one). Once DM-828 is done, we should put together a Confluence page describing the design, and email lsst-data for a broader design review (I hope we can do that by email, but we'll try to schedule a live session if we have to).

          People

            jbosch Jim Bosch
            jbosch Jim Bosch
            Perry Gee
            Jim Bosch, Perry Gee
            Votes:
            0 Vote for this issue
            Watchers:
            2 Start watching this issue

            Dates

              Created:
              Updated:
              Resolved:

              Jenkins

                No builds found.