Maohai, I really need a test program. Unfortunately, this problem is probably only vaguely related to msg16052.html in the visad-list archive. I think the basic problem is that the call to createImage() in the VisADCanvasJ2D constructor is retuning null when you add it to an internal frame, and the calls to getSize() are returning width = height = 0 (we fudge them to 1). However, my current fixes for these conditions don't seem to be working. So I need a test program. Cheers, Bill On Sat, 1 Mar 2003, m huang wrote: > Bill, > > I have tried the new jar in untested. the problem is > still there. I have found out these: > > The problem of disappearing box/plot only manifests in the > first plot in the application I am working on, if I open > multiple plots. Although I have seen it happening in the > second plot in a test program. > > If I open a test display that has domain and range all set up > x range [-1, 10], yrange [-2.3, 9], after I drag the mouse on it to > make the plot disappear, I click both the right and left > mouse buttons to simulate the center button. The cursor > coordinates still show on the upper left corner, but the > numbers are wrong (in the range of hunreds and thousands). > This leads me to drag the plot "by instrument" so that > the orginal box could appear in the display area according to > the displayed coordinates. I find that with very small > drag of the mouse, the coordinates change a lot. But finally I > center the plot. Yes the plot is still there. Somehow > it got rescaled to about 5 or 6 times larger and is simply > span out of the internal frame window when the mouse is dragged on > it for the first time. The confusion is perhaps related to > a (undesireable) visad mouse behavior reported here: > > http://www.unidata.ucar.edu/staff/russ/visad/msg16052.html > > > hope this helps.