Search All of the Math Forum:
Views expressed in these public forums are not endorsed by
NCTM or The Math Forum.



Bug in PlotBoxAspectRatio
Posted:
Jun 19, 1997 12:47 AM


I initially reported an example where using the recommended procedure with DAR and PBAR both MANUAL and both set to simple [1,1,1], the resulting axis limits are automatically calculated to be so large that the data is only visible as dots at the center of the figure. In other words, the axis limits are set too large. The confirmation of the bug from Rick was as follows: > Yes I do see your problem. I think there may be a bug in the > limits calculation when both PBAR and DAR are in MANUAL. The bug > being that the limits computed are not the tightest that could be used. >
I do not understand the subsequent retraction, nor the justification that
> the PlotBox indeed takes the aspect ratio asked for but the data now > collapses to a sliver. For some cases that is the right answer; let
The justification proceeds to use an example where the data ratios are 100 to 1 or even 1,000,000 to 1 in y to x ratios. However, this is not at all the case. My particular data are entirely reasonable without any such extreme ratios. In fact, the ratios are approximately 1 to 1. For an actual example, see the first color figure in the section of my web page at
http://www.toolsmiths.com/firwav.shtml#indfil
where I am unable to get the plotboxes in the left column of subplots to have the same shape and size as the other plotboxes. Yet the data is entirely reasonable being nothing more than the unit disk in the complex Z plane. Here in order for the unit disk to appear as a circle it is necessary for the DataAspectRatio to be [1,1,1]. So once again, I cannot accept the justification for the actual behavior, and I do believe that it is either a bug or a design flaw.
With regard to the rest of the discussion, I also maintain the points I made in my previous post:
Handle Graphics should provide control to the user to enable design and display of plots and graphics without inexplicable or nonintuitive hidden behaviors or otherwise complicated sequences of interlinked and interacting effects for which the rules of interaction are not documented. If I change a given property, I should be able to expect a given result in appearance that is in fact reflected in a change in the numerical value corresponding to the appearance of the feature. When the appearance can change WITHOUT a corresponding change in a numerical value that can be controlled with a specified property, then obviously the user/developer/programmer NO longer has control over his programming and/or graphic design.



