[Octave-bug-tracker] [bug #45292] Strange issue with image.m and Gnuplot

[Alexander Klein]

URL:
  <http://savannah.gnu.org/bugs/?45292>

                 Summary: Strange issue with image.m and Gnuplot
                 Project: GNU Octave
            Submitted by: matalex
            Submitted on: Mi 10 Jun 2015 11:03:36 GMT
                Category: Plotting with gnuplot
                Severity: 3 - Normal
                Priority: 5 - Normal
              Item Group: Regression
                  Status: None
             Assigned to: None
         Originator Name: 
        Originator Email: 
             Open/Closed: Open
         Discussion Lock: Any
                 Release: 3.8.2
        Operating System: *BSD

    _______________________________________________________

Details:

When I tried to use the "image" command to display a matrix with Gnuplot, I
got the following two warnings:

1. operator -: automatic broadcasting operation applied, (this was shown
twice)
2. image: non-linear X, Y data is ignored. IMG will be shown for linear
mapping.

I haven't investigated the further, but the latter is obviously incorrect for
my set of coordinates, and it turned out that the whole process of determining
whether the input coordinates are linear is probably confused by meshgrid
matrices being used as an input to image.m, and as a consequence, numel
returning an unexpected result in the original version of lines 166 and 169.

The whole problem could be patched away by the following changes to image.m,
but may warrant a bit of further investigation by someone who is more familiar
with the respective code:


169c
      dymean = (max (y) - min (y)) / (rows (y) - 1);
.
166c
      dxmean = (max (x) - min (x)) / (rows (x) - 1);
.





    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/bugs/?45292>

_______________________________________________
  Nachricht gesendet von/durch Savannah
  http://savannah.gnu.org/