[joernu76]

The takeoff time could only be set in the Performance Settings dock, which few users found. The Time (UTC) column of the table view (shown with "Show Performance") is now editable with a date/time editor with calendar popup. Setting the time of a waypoint sets the takeoff time so that this waypoint is reached at the given time; all other times follow from the aircraft model. The takeoff time field is removed from the dock and replaced by a hint.

• All times are handled in UTC. QDateTimes are built directly in UTC instead of via local time, which gave invalid values for times in a daylight saving gap of the local zone (the editor opened at 1999-12-31 and shifted the flight there).
• Time (UTC), leg time and cumulative time are rounded to full seconds instead of truncated, so an entered 12:00:00 no longer shows 11:59:59.
• The top view redraws on time changes (solar angle lines).
• WaypointsTableModel.load_settings passes a copy of DEFAULT_PERFORMANCE: load_settings_qsettings updates its defaults in place, so all models shared, and modified, the module-level default dict.
• The take off time step is removed from the performance tutorial and its texts; docs/usage.rst describes the editable time.

Purpose of PR?:

Fixes #3316

Does this PR introduce a breaking change?
No

[ReimarBauer]

Hi @joernu76

my standard code-review high by claude shows hints for some unwanted options the changes create. Please have a look into these

Likely real bugs:

1. Leaving the Time (UTC) editor without a change still shifts the flight (flighttrack.py:984). The editor shows whole seconds and writes that value back. Example: a waypoint at 13:00:00.6 shows 13:00:01; committing it moves takeoff to 11:59:59.4. The other waypoints can then show times one second off. Fix: skip setData when the value is unchanged.
2. Times saved by earlier versions are still in local time (flighttrack.py:786). The fix only applies to newly edited times. A takeoff time saved by the old dock widget was stored in local time and keeps showing as UTC, or wrong around daylight saving changes, until the user edits a cell. Fix: convert the loaded takeoff_time to UTC once, in load_settings.
3. Only an editor that sends UTC gets a correct time (flighttrack.py:660). setData uses value.toPyDateTime() without toUTC(). Any caller passing a time in local spec gets it taken as UTC. value.toUTC().toPyDateTime() works for every spec.
4. The solar-angle time can no longer be set with performance off (performance_settings.py:91). Removing the dock's takeoff field means the takeoff time can only be set with Show Performance ticked. The top view's solar-angle lines still depend on it, so a user can't change their time without switching performance on.

Performance and side effects:

5. Extra redraws (mpl_pathinteractor.py:1331). The top view now listens to TIME_UTC, which update_distances emits on every change. So every edit in the table redraws the top view twice, and inserting or removing rows also redraws in the middle of the operation.
6. MSColab sees a change that isn't one (flighttrack.py:669). A time edit emits dataChanged, which MSColab treats as a flight track change. Working locally, the FTML file is written twice; online, right after a load, the unchanged content may be sent to the server.
7. Full recalculation for a time shift (flighttrack.py:670). A time edit reruns the whole performance model for every leg, although shifting each waypoint's time by the takeoff difference would be enough.

Small:

8. The new hint label in ui_performance_dockwidget.ui has no wordWrap, so it makes the dock several hundred pixels wide.
9. test_edit_time_in_local_dst_gap calls time.tzset(), which exists only on Unix. Skip the test where it's missing.

[joernu76]

I have to talk with my Claude Code Review high then...

[joernu76]

OK; the code was not very well thought out to begin with. I asked for a clean rewrite and check of updates, etc. Tests are much more comprehensive and I believe that we fixed several minor bugs on the way.