Re: Stem update, rough guess.

[Daniel J Sebald]

John W. Eaton wrote:
> On  9-Apr-2007, Daniel J Sebald wrote:
>
> | The baseline that is a child of the axis seems like what I had in mind, i.e., 
> | zero axis line, i.e., something very general that applies to not only stem 
> | plots.
>
> Yes, but that child is just a simple line.  There seems to be nothing
> special about it that indicates that it is a "baseline".
>
> | Does stem allow a baseline other than at y=0?
>
> Yes,
>
>   h = stem (-10:10);
>   set (h, 'basevalue', 3);
>
> works.
>
> | If so, that would explain the extra baseline handle.
>
> | I'd say, leave out the one that is passed back as a 
> | handle and implement the one that is a child of the axis using the gnuplot 
> | command.
>
> Yes, I think this would be OK, but again, the line that is a child of
> the axes object does not appear to be special.
>
> Again, the X data for both the child of the axes object and the
> baseline object that is a child of the stemseries object changes when
> the axis limits are changed.  So it is not sufficient to have a
> special property that is recognized and translated into a special
> gnuplot command.  Instead, the data for these objects should be
> modified when the limits change and then drawing the lines will
> generate the proper plot.  But since there is probably no way to do
> this completely right until we have group objects, I think it would be
> OK to have the baseline be a child of the axes object and not worry
> about resizing it automatically.  I think that is a small bug compared
> to not having it as a line object.

Well, that's kind of strange then.  The creation of the baseline can't be in
stem.m using "line()" because the "set" of

>   h = stem (-10:10);
>   set (h, 'basevalue', 3);

comes after the stem.m.  stem.m would have to pass a handle to set.m somehow and
that would make no sense.

Of course, inside stem.m there can be a "set (h, 'basevalue', 0)" as default.

Well, in any case the attached patch will add back in the baseline.  For the 
same logic of allowing a baseline unadjustable for now, I've also made the 
baseline unconditional rather than conditioning on, say,

  if (find (y < 0) && find (y > 0))

so that if the y range is expanded, the baseline will be visible.  For example, 
try

 h = stem(0:10);
 set(gca,"xlim",[-1 12]);
 set(gca,"ylim",[-1 12]);

after the patch.

Dan

--- /usr/local/src/octave-cvs/octave/scripts/ChangeLog  2007-04-09 
19:46:52.000000000 -0500
+++ ./ChangeLog 2007-04-09 20:15:08.320764394 -0500
@@ -1,3 +1,8 @@
+2007-04-09  Daniel J. Sebald  <address@hidden>
+
+       * plot/stem.m: Add back the baseline line and let it be not adjustable
+       in response to a change in x limits for now.
+
 2007-04-09  John W. Eaton  <address@hidden>
 
        * plot/__go_draw_axes__.m (__gnuplot_write_data__): New function.
--- /usr/local/src/octave-cvs/octave/scripts/plot/stem.m        2007-04-09 
14:45:26.000000000 -0500
+++ ./stem.m    2007-04-09 20:26:53.981246122 -0500
@@ -113,12 +113,17 @@
   z = zeros (1, numel (x));
   xt = x(:)';
   yt = y(:)';
-  tmp = plot ([xt; xt], [z; yt], "color", lc, "linestyle", ls,
-             x, y, "color", mc, "marker", ms, "linestyle", "",
-             "markerfacecolor", fc);
+  h_stems = plot ([xt; xt], [z; yt], "color", lc, "linestyle", ls,
+                 x, y, "color", mc, "marker", ms, "linestyle", "",
+                 "markerfacecolor", fc);
+
+  ## Must draw the plot first to get proper x limits.
+  drawnow();
+  x_axis_range = get (gca, "xlim");
+  h_baseline = line (x_axis_range, [0, 0], "color", [0, 0, 0]);
 
   if (nargout > 0)
-    h = tmp;
+    h = [h_stems; h_baseline];
   endif
 
 endfunction