Bizarre figure_size convention in 5.1.1

[StŽéphane Mottelet]

Hi all,

this as been bugzilled #4521, but I discovered that really big problems are
a consequence of this issue. I am currently developping a feature
which needs to identify the handle of an axis where a mouse
clik occured (with an event_handler function). Since the coordinates
recovered in the event handler function are integer coordinates in
the *real*canvas of the figure, it is impossible to compute the
real limits of the axis in pixel coordinates (using the axis_bounds
property), since the figure_size property is wrong (the height is
not equal to the height of the canvas).

I now there may be more critical bugs, but this one is really
problematic for people developping some stuff related
to graphics...

S.

StŽéphane Mottelet a écrit :
> Hi again,
>
> seems that the figure_size definition has changed in 5.1.1, this 
> causes some
> different behaviour between e.g. 4.1.2 and scilab>5.
>
> In Scilab 5.1.1, at least under windows xp, the height of the figure 
> given
> by get(gcf(), figure_size)  is equal to the full height of the window, 
> taking
> into account the height  of the scilab toolbar and the height of the 
> window
> manager's window bar.
>
> In Scilab 4.1.2, the height given by figure_size is equal to the 
> height of
> the canvas + the height of the Scilab toolbar.
>
> In Matlab, the height of the window given by get(gcf,'Position') is 
> equal to
> the height of the canvas, which seems to be the most logical option (i.e.
> platform independent).
>
> S.
>