GitLab

Fix behavior of DBus method setRaDec, provide a new DBus method setRaDecJ2000, and support refraction-corrected altitudes in DBus method setAltAz

This commit does three things:

1. Fix a mysterious behavior in setRaDec, which "overcorrected" for
   atmospheric refraction in some buggy way. Unlike
   SkyMap::setDestination, SkyMap::setClickedPoint does not seem to
   suffer this atmospheric refraction issue -- so using
   SkyMap::setClickedPoint instead of SkyMap::setDestination in the
   DBus method setRaDec and then calling SkyMap::slotCenter works! We
   largely use setClickedPoint to set the center of the star map
   almost everywhere across KStars (if my memory serves me right), so
   this is not a bad workaround.

   KStars::setRaDec now replicates the behavior of typing the same
   RA/Dec values in the Focus Dialog, as far as I can tell.

   However, this does not fix the behavior of SkyMap::setDestination,
   and why SkyMap::setDestination has this behavior is still to be
   investigated.

2. A similar change to the above is also applied to DBus method
   org.kde.kstars.setAltAz, whereby it now replicates the behavior of
   the focus dialog.

3. The behavior of the focus dialog ("Set Coordinates Manually") in
   setting Alt/Az coordinates manually is to assume that the supplied
   altitude is the true (i.e. before refraction) altitude. Which means
   that, if you enter an altitude of e.g. 5 degrees, the position
   centered will actually have apparent altitude > 5 degrees when
   refraction corrections are enabled.

   This same behavior is reflected in org.kde.kstars.setAltAz
   now. However, there may be reason to center using the refracted
   coordinates instead. For that, we now support an optional bool
   argument altIsRefracted to the DBus interface method
   KStars::setAltAz, that when set to true, will "do the best we can"
   to interpret the given altitude as the _apparent_ altitude of the
   center.

Finally, a bit of an administrative / logistical note: This is my
first commit after a long time. I am no longer a PhD student, and am
employed in California. Unlike in many other parts of the US,
California state law allows me to hold copyright on code written in my
personal time and on my personal equipment, as long as there is no
conflict of interest with the goals of my employment. I wish to indeed
explicitly clarify that these and future contributions that I make to
KStars are made in my personal capacity, with my personally-owned
equipment and personally-paid internet connection, and I am not
conveying any rights to any intellectual property of any third
parties.

Summary:
When opening the mount control box (button top right corner of the Mount tab), some elements are partially hidden.

Before: {F8230732}

After:{F8230737}

Reviewers: mutlaqja

Reviewed By: mutlaqja

Subscribers: kde-edu

Tags: #kde_edu

Differential Revision: https://phabricator.kde.org/D28735

Summary: Previous submit needs an include of <cmath> to compile on Macs.

Test Plan: Compile kstars on a mac, won't work without this.

Reviewers: mutlaqja, TallFurryMan

Reviewed By: mutlaqja

Subscribers: kde-edu

Tags: #kde_edu

Differential Revision: https://phabricator.kde.org/D28716

Summary:
This differential is a hotfix for the recent work in the Focus module.
One additional class, FITSSkyObject, was extracted and put in its own source files.
The JMIndex parameter of the Centroid detection was not provided to the source extraction procedure.
The Star Profile Viewer missed a few consts in its management of FITS data buffer, which caused some Jenkins builds to fail on SUSE and FreeBSD.

Test Plan: Run kstars_ui_tests

Reviewers: #kstars, mutlaqja

Reviewed By: #kstars, mutlaqja

Subscribers: kde-edu, mutlaqja, murveit

Tags: #kde_edu

Differential Revision: https://phabricator.kde.org/D28689

Summary:
This differential adds a few FITS load tests and benchmarks.
This alone is not enough to cover the whole detection algorithms, but is a beginning.

Test Plan: Build and run "testfitsviewer"

Reviewers: #kstars, mutlaqja

Reviewed By: #kstars, mutlaqja

Subscribers: murveit, kde-edu

Tags: #kde_edu

Differential Revision: https://phabricator.kde.org/D28635

Summary:
This differential relates to tests and code structure for the Focus module.
Tests show that:
- There is no way to determine when the star detection procedure is done. The capture button should re-enable when everything is finished.
- The CCD Simulator needs improvement for some detection mechanisms to work properly.
- It is difficult to create fixtures with deterministic verifications, although better controlling the simulation time should help.
- Syncing the mount to object coordinates is a good idea to speed up tests.
- When not tracking, the CCD Simulator doesn't render trails, it could be a nice improvement.
- Because kstars_ui_tests is becoming larger, it needs to be divided into multiple smaller tests for efficiency.
- QTest, because failing a test means returning from the test function, is macro hell.
- QComboBox is difficult to control programmatically because of its internal signal management.

We then refactor star detection for clarity and consistency in Focus and attempt to reduce side-effects in FITSData:
- Make FITSData buffer read-only, except for DarkLibrary.
- Move Threshold, Gradient and Centroid detections out of FITSData.
- Move SEP detection out of FITSData.
- Docs, copyrights and adjustments.
- Removing expected failure on Gradient/Threshold.
- Adjust KTRY_EKOS_CLICK to use KTRY_EKOS_GADGET.
- Extract star search block to simplify setCaptureComplete.
- Reset KStars time after starting Ekos.
- Fix object fixture builder.
- Slight delay to overcome the weird toggling of the capture button.
- Separate/simplify annulus and threshold tests.
- Delay target chip retrieval when capture is complete.
- Test frame average, extract code averaging HFRs.

Test Plan: Run kstars_ui_tests.

Reviewers: #kstars, mutlaqja

Reviewed By: #kstars, mutlaqja

Subscribers: murveit, mutlaqja, kde-edu

Tags: #kde_edu

Differential Revision: https://phabricator.kde.org/D28292

Reviewers: mutlaqja

Reviewed By: mutlaqja

Subscribers: kde-edu

Tags: #kde_edu

Differential Revision: https://phabricator.kde.org/D28294

removeProperty is called in the BaseClient thread and provides a pointer to INDI::Property object. Since INDI_D::removeProperty method needs to run in the main GUI thread, we need to invoke the method in the primary thread. However, blocking the calling thread to do so might result in deadlock. Using QueueConnection solves one issue, but by the time removeProperty is invoked, the INDI::Property object could have been deleted already resulting in a crash.

The current solution uses the name of the property to invoke the method in the primary thread using QueuedConnection, however, there is no pointer to lose now since we just send the name.

BUGS:416888
FIXED-IN:3.4.2

BUGS:416173
BUGS:398192
FIXED-IN:3.4.2

Summary: Fixes Ubuntu 16.04 test reported as https://launchpadlibrarian.net/469832897/buildlog_ubuntu-xenial-amd64.kstars-bleeding_6%3A3.4.2+202003200441~ubuntu16.04.1_BUILDING.txt.gz

Test Plan: Launchpad build.

Reviewers: mutlaqja

Reviewed By: mutlaqja

Subscribers: kde-edu

Tags: #kde_edu

Differential Revision: https://phabricator.kde.org/D28177

Summary:
Restricts fixture-based test to QT>5.9.
Removes initial condition test on kstars date for QT<5.8.
This is a no-brain workaround to return build on Ubuntu 16.04 to stable.

Test Plan: Local build OK, check Launchpad build.

Reviewers: mutlaqja

Reviewed By: mutlaqja

Subscribers: kde-edu

Tags: #kde_edu

Differential Revision: https://phabricator.kde.org/D28117