[Celestia-developers] goto-Bugfix, Mouse-usability

[Harald S. <har...@us...>]

Hi,

three independent items in this mail, one bugfix and to suggestions
about mouse-control.

First, there is a bug in Observer::computeGotoParameters(), which is
triggered when celestia is in a "locked" frame (for example: "lock
earth-sol") and then goto to the target-object (here: sol) is
activated. A new FrameOfReference with coordSys PhaseLock but without
the required targetObject is set, which leads to a "warped" (Linux) or
a black (Windows) screen. The same bug seems to be in
computeGotoParametersGC(), but I am not sure if or how it can be
triggered.

Solution: either don't touch the frame-of-reference at all if using
PhaseLock, or explicitely set it to some other mode (follow?).

For the first fix, replace in line 528 (and possibly in 593) of
observer.cpp:

    setFrame(FrameOfReference(frame.coordSys, destination));

with 

    if (frame.coordSys != astro::PhaseLock)
    {
        setFrame(FrameOfReference(frame.coordSys, destination));
    }

======== 

Second: Rotating the observer around a planet with right mousebutton
while very close to the surface is IMHO much to sensitive, which
becomes important with some Virtual Textures (like the Kennedy Space
Center-texture). I made a change for CelestiaCore::mouseMove which
computes the altitude above the surface for stars and planets and if
it is less than the radius the coarseness-factor is multiplied by
(alitutude/radius) - this worked pretty good for me (may need some
tuning for general taste regarding rotation speed).

========

Third & last: I personally would prefer to have the rotation along the
observer-axis (roll? i.e. what is controlled with Keypad 7 & 9)
controllable by mouse too. Currently, pressing Left & Right
mousebutton with horizontal movements does nothing and can be used for
this. One problem with this idea is accidental rotation when changing
distance with movements along the vertical axis. To avoid this, I
tried using a condition to only rotate when movement along horizontal
axis is larger than along the vertical axis. Some people may still not
like this, but as for example moving around a planet does easily change
the orientation relative to the ecliptic or the equator, many users
probably need to readjust it. 


Chris, I think these changes are small enough to still go in 1.3.1. If
you want to check-in (or just test) any of these changes, just tell me
which ones and I will mail you the accordingly changed files.

 Harald