Skip to content

Fix the focus state notification

Eric Dejouhanet requested to merge edejouhanet/kstars:bugfix__focus_state_report into master

This MR introduces tests to validate the notification of the focus state to other modules, and refactors the completion of the notification.

It resolves the problem with the order of calls when completing a focus procedure, which was allowing client modules to restart a focus procedure while the previous was not finished yet.

Two issues are raised by these tests, located in the CCD Simulator:

  • The default rotator randomisation does cause strong non-realistic jitter when looking at the celestial pole.
  • The amount of light is miscalculated in the gaussian star shape, causing stars to disappear into a single dot at optimal focus and stretch too bright when out of focus.

These issues make the tests more complicated to consolidate. Star detection is really difficult when approaching optimal focus and the focus procedure is thus bound to fail for certain algorithms because of noise levels.

Note that this MR does not fix the Mount/Focus conflict happening when the meridian flip happens during a focus procedure. It fixes the race condition between Scheduler and Focus, which happens when Focus fails and Scheduler immediately asks for a new autofocus.

Edited by Eric Dejouhanet

Merge request reports