GitLab

I was working on the Whats Interesting tool because I think it has a lot of potential.  I added photos and descriptions to the list.  I added photos to the detail view.  I also changed some of the layouts.  And finally I added a telescope slew button.

Fix a couple of issues with flags. They are supposed to be saved in J2000 internally but where in JNow before.
Fixed other issues with adding, updating, and removing flags.

CCMAIL:kstars-devel@kde.org

Use dark image in all FITSView widgets until an actual image is captured. Same image is also used in Observation Planner to denote no image

This is necessary ever since we changed the Observation Planner to use
clone SkyObjects instead of the versions in the map.

TODO: It might be better to make showEyepieceField take a const
      SkyPoint and clone it.

TODO: It would be good to show the time, date and geolocation of the
      eyepiece field view.

1. Initialize DSS image metadata correctly

2. Add a bool to indicate whether metadata is valid or not

=> These two fixes prevent KStars from thinking that the image has
   metadata in cases where it does not actually have any, but there is
   junk initialization values.

3. Use the isValid() check which is neater.

Clearly, oversight!

Fixing downloader issue and showing progress dialog where appropiate. It starts with unknown progress and then once progress is established it starts to show it. Fixing imageview, it should not be shown until image is ready

These were presumably caused because of the use of clones of the
original SkyObjects in ObservingList (see
f11601a3). These cloned SkyObjects
don't have updateCoords or EquatorialToHorizontal called on them.

This was causing two problems:

1. The Sky map "screenshot" would not center correctly on the object,
   but would instead always be off by a few arcminutes.

2. The DSS image would have an indeterminate rotation angle and
   thereby would not be rendered at all.

This patch fixes these two issues, but has potential for introducing
regressions in the Eyepiece View as invoked from the sky map
(KSPopupMenu) or in the export option with ticks.

Added functionality for Update button. Now, the AltvsTime widget will recompute all the graphs points when the Update button is pressed.

a.k.a. the "Wishlist-Killer" bug.

A very frustrating crash would ensue whenever the catalog
configuration options were changed (eg: adding a new custom catalog)
and an object was added to the observing wishlist. Before a recent
commit, the wishlist would be truncated causing much frustration to
the user.

The crash occurred because the SkyObjects are de-allocated and
re-allocated by SkyMapComposite::reloadDeepSky() every time the
catalog configuration objects have changed. Since the ObservingList
maintained a QList<SkyObject *> pointers to these SkyObjects, we were
often querying dangling pointers, leading to crashes. So addition of
the object to the wishlist, although did trigger the bug, was not the
root cause.

This change fixes the bug (hopefully without inducing any
others!). The changes are:

1. Create a clone of the SkyObject in ObservingList

   This makes add / remove more complicated as they look-up the given
   object by name, but solves the crash because we have a local copy
   of the SkyObject even if the CatalogComponent's copy is deleted.

2. Use QSharedPointer<SkyObject> instead of SkyObject *

   This makes sure that we don't re-create the same object for session
   plan / wish list. Besides, it is much more cleaner, and takes care
   of deleting itself etc.

Extensive testing is appreciated.

And once again, happy 20th Birthday KDE!

BUG: 345348
CCMAIL: kstars-devel@kde.org

There is a bug which induces a crash when writing the observing wish
list, and ends up truncating the wish list. This can be very
frustrating because of the loss of the wishlist.

Although the bug still needs to be found and fixed, this change in
code at least prevents the wishlist from being truncated upon crash.

Disabling a few calls uncompatible with the new QCustomPlot. Need to update it to use QCPAxisTickerTime later on

KSDssDownloader commits suicide, so code like:

KSDssDownloader *foo = new KSDssDownloader( ... );
connect( foo, ... );

has potential to fail.

N.B. The better and more correct fix would be to make the one-shot
use-case of KSDssDownloader a static method, just like the one-shot
QTimer. It can internally create a class instance and safely handle
its destruction.

Must use deleteLater() as there may still be events to process

This seems to fix the crash (at least I cannot reproduce it as easily
any more)

It seems more natural to provide more vertical real estate for the
Table View objects, since the list spans a fairly long length in
typical cases. Just showing 4 rows is not very useful, but showing as
few as 2 columns should still be useful (it takes less scrolling to
scroll through the small number of columns).

However, we would want to make the first column, the one with the
names of objects, always visible irrespective of how we scroll
horizontally. To this end, there is a solution in the form of a Qt
example which we can use in KStars with minimal modification:

http://doc.qt.io/qt-5/qtwidgets-itemviews-frozencolumn-example.html

N.B. The BSD License used in the example is GPL-compatible.

Other improvements necessary are:

1. Put the list on the left -- feels more natural.

2. Move the time schedule option in the Session View below the table
   view.

3. Gray out the date box in WishList mode to indicate its irrelevance?
   It is technically not irrelevant since the RA/Dec shown are
   apparent coords! Maybe this should change.

4. Show some vital info about the object in a label

5. Make sure buttons don't expand if we resize -- only the table view
   must expand to fill up as much real estate as possible.

6. Allow the user to hide / show columns in the TableView.

7. Make sure the "Minify" option works well.

8. Would be nice if we could "minimize" the observing list to an icon
   in the bottom-right of the main window.