Details
-
Type:
Improvement
-
Status: Invalid
-
Resolution: Done
-
Fix Version/s: None
-
Component/s: pipe_tasks
-
Labels:None
-
Epic Link:
-
Team:Data Release Production
Description
For processing data which already has a good WCS (such as CFHT and SDSS) it is unnecessary to perform an initial fit of astrometry before computing the PSF. Furthermore it is risky to do so because the astrometry fitter may fail.
At present CFHT handles this by overriding the CalibrateTask with its own variant task and a variant run method. However, the amount of code duplication required is unacceptable.
To avoid this I suggest we add the ability to disable the initial astrometric fit.
I also think we should Refactor CalibrateTask so its run method is much simpler and easier to override. But even if we do that (and perhaps it should be a separate ticket) it is such a common need to not run the initial astrometry that I think one should not have to override the run method to do it.
Attachments
Issue Links
- duplicates
-
DM-4692 Refactor ProcessCcdTask and sub-tasks
- Done
Activity
| Field | Original Value | New Value |
|---|---|---|
| Description |
For processing data which already has a good WCS (such as CFHT and SDSS) it is unnecessary to perform an initial fit of astrometry before computing the PSF. Furthermore it is risky to do so because the astrometry fitter may fail.
At present CFHT handles this by overriding the CalibrateTask with its own variant task and a variant run method. However, the amount of code duplication required is unacceptable. To avoid this I suggest we add the ability to disable the initial astrometric fit. I also think we should Refactor CalibrateTask so its run method is much simpler and easier to override. But even if we do that (and perhaps it should be a separate ticket) it is such a common need to not run the initial astrometry that I think one should not have to override the run method to do it. Some thoughts on how to disable the initial astrometric fit: - One obvious choice is to add a calibration parameter to control this. It would then be overridden in the appropriate obs_* package. This provides flexibility, which might be useful for testing, but risks that the parameter will be mis-set (e.g. if CalibrateTask is used by more than one task then additional overrides become necessary). - One could imagine more hard-coded designs where the setting is fixed for a given obs_ package, e.g. by adding a class variable flag or constructor parameter. |
For processing data which already has a good WCS (such as CFHT and SDSS) it is unnecessary to perform an initial fit of astrometry before computing the PSF. Furthermore it is risky to do so because the astrometry fitter may fail.
At present CFHT handles this by overriding the CalibrateTask with its own variant task and a variant run method. However, the amount of code duplication required is unacceptable. To avoid this I suggest we add the ability to disable the initial astrometric fit. I also think we should Refactor CalibrateTask so its run method is much simpler and easier to override. But even if we do that (and perhaps it should be a separate ticket) it is such a common need to not run the initial astrometry that I think one should not have to override the run method to do it. |
| Link | This issue relates to DLP-495 [ DLP-495 ] |
| Team | Data Release Production [ 10301 ] |
| Epic Link |
|
| Resolution | Done [ 10000 ] | |
| Status | To Do [ 10001 ] | Invalid [ 11005 ] |