[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

20051107: IDV: Bugs under Mac OS X?


Hi Dave,



For the last several weeks I've been muddling through my first attempt to
teach some of our meteorology majors how to use the IDV, using the Mac G4
laptops running OS X (v10.9.3, Panther) in our computer labs. We've
stumbled across several anomalies that I'll call "bugs", though whether
these bugs are in the IDV, in Apple's implementation of Java, in the
interaction between the two, or in the users (us), I'll let you decide.

  (1) In the View menu, when we select "Full Screen", we get a full
screen all right, but in most cases we don't get the menu bar across the
top of the display window. In the one exception I've seen (a student's
privately owned Mac G4 laptop, also running OS X v10.3.9, I think) the
upper left corner of that menu bar has a button that returns the view to
non-Full Screen view, but on our machines we don't see that. I have a Mac
Cube running OS X 10.4.3 (Tiger) in my office that shows most of a menu
bar in Full Screen mode, but nothing happens when I click on the spot
where I think button should be.
  We discovered by trial and error that when we're in Full Screen mode,
pressing <Option>V pulls down the View menu, but clicking on the "Full
Screen" item in that menu doesn't toggle Full Screen mode back off. (This
is what I think should be called a bug--all the other toggle items in the
View menu have a check mark beside them when they're on, and they truly do
toggle something on and off.) The same is true in the couple of instances
(the student's Mac G4 laptop and my Mac Cube) where we can actually see
the View menu in Full Screen mode--that is, we can pull down the View menu
and click on "Full Screen", but nothing happens
The lack of toggling of full screen mode is a bug that we will fix. I would say the rest is Mac/Java weirdness. Basically, we just make w very big window that is placed at (0,0) screen coordinates (upper 
left). We will try to do something that is "Mac" sensitive.


  (2) When we have two separate displays open, and one display has two
panes and the other has one pane, and "Shared Views" is turned on on all
displays, the two separate displays don't share rotation, panning, and
zooming (though they do share animation controls). On the other hand, the
two panes in the two-pane display do share these controls.
That's not quite a bug but a feature. There are sort of two classes of things that are being shared here - the viewpoint state and the time animation. By default, all time animation is shared, irrespective of the windows. However, when you create a 2 pane display the display state is only shared between those two displays. 
The sharing is "group" based. All a group is is a name (e.g., "Animation").
For the linked multi display windows the group is a big unique number.
In the latest development release we provide the user the ability to change the 
share group in a display and in an animation. For example,
to address your above problem you could change the name of the group
in each of the displays to some group name, .e.g. "Group1".
You do this through the Projections->Set Share Group menu.


  (3) The mouse-over identifier labels on the Viewpoint Toolbar and
Navigation Toolbar (that is, the labels that appear next to an icon when
the mouse lingers on the icon) should appear on top of the display window
frame and the (normally) black display window so that we can read the
mouse-over label. They do appear on top of the display window frame but
the portion of the label that should lie on top of the display window
itself is hidden, as if it lies beneath the display window. As a result we
can see only part of the identifier label.
This sound like a Java heavyweight vs. lightweight component problem. The 3D display component is heavyweight and does not play well with lightweight components. On Linux the tooltip is not a problem. We will note this but I'm not sure if we can engineer a solution for it. 


  (4) When we plot an isosurface of a 3D field (say, temperature or wind
speed) from a NAM model forecast file (say), and decide to change the
isosurface value, in the display control window we enter a new value in
the isosurface value entry box and press the return key. As a result, the
value entered there is decremented by 1 (or incremented by 1 if the value
is negative) and the isosurface is then plotted with the decremented
value. (I don't see any other way to signal to the IDV to replot with the
new isosurface value.)
That is indeed a bug and we'll fix it. You can change the value by dragging the slider. 
You could also enter a value 1 greater than you want :-)


  (5) In general, for single-button mice (or click pads on Mac G4
laptops, for example), the right button on a two- or three-button mouse is
simulated by holding down the <Option> key and clicking with the single
button [<Option>click]. In the IDV, this works for zooming
[<Shift><Option>click] and rotating [<Option>click] a display. However,
with a two- or three-button mouse, panning is achieved with
<Cntl>right-click. I would have thought that the equivalent on our
one-button laptops would be <Cntl><Option>click, but this doesn't
work--the cursor can be moved but the display doesn't follow.
Not being Mac people (i.e., Don and I) we haven't had a chance to deal with many of these issues so please bear with us. We have had issues where a Control-click is interpreted 
as some other kind of mouse click which is probably what is happening here.
We will look into this but, short of a revamp of our key/mouse click functions, 
there might not be much we can do.



  The panning icons on the Navigation Toolbar produce large panning
jumps, and while the arrow keys produce smaller jumps those jumps are
still far from smooth. Hence there seems to be no way to produce a smooth
pan.

Good point. We will make the panning arrows use smaller increments.


These are relatively small problems, and the students are for the most
part blown away by what they can do with the IDV. One student, who teaches
our one-unit non-majors intro lab course for us, has immediately begun
integrating some of what she learns about the IDV into the lab exercises
that she designs each week.


Great. We always appreciate hearing good news.

-Jeff