Skip to content

Internal Guider fixes

Hy Murveit requested to merge murveit/kstars:guider-fix1 into master

Overview

There are three internal guider issues fixed in this MR.

  1. SEP MultiStar has been unconstrained as to where it locates the guide star. With this change, it must either find the guide star in the reticle box or fail. The user can enlarge the reticle, as before.
  2. Dithering previously was done relative to the star position. It is now done relative to the reticle. When the star and reticle are close, as is typically the case, then this is not much different. However, there were unusual times when (see issue #1 (closed)) SEP MultiStar detected a guide star far away from the reticle right before a dither. This caused a large jump. 
  3. When the scheduler re-starts the guider, previously the GuideStars algorithm was not reset, and thus it did not re-initialize it's reference stars. This is now fixed.

Details

gmath.cpp/h

  • added abort() method which resets GuideStars (fix 3 above)

guidestars.cpp/h

  • reinitializes the star correspondence if findGuideStars() is called but starCorrespondence is uninitilized.  (3)
  • Add logging if findGuideStars fails.
  • Requires findGuideStars solution to be in the tracking boxAdd reset method that resets starCorrespondence (1)

internalguider.cpp

  • Adds call to gmath::abort() in its abort() method (3)
  • Changes dithering to be relative to the reticle. (2)

fitssepdetector.cpp

  • modifies an existing log message to add columns related to sky background.

Other Notes

This is part of debugging logs generated by @alacant. This seems to fix his issues.

The logging added should have minimal impact. In one case it's adding columns to an existing debug log, in the the other it only prints on a failure, which should rarely happen and facilitate debugging.

Testing

Difficult to set up this situation in the simulator, as these fixes really address issues to make the internal guider more robust when the guider fails to properly find the guide star, or if something unusual causes the guide star to move in the guide image. I should add test cases to cover this, but not in the current MR.

Edited by Hy Murveit

Merge request reports