[Kst] on plots and windows

[C. Barth Netterfield]

On Saturday 10 April 2004 6:22 pm, George Staikos wrote:
>    What is desired for the behaviour of plots and windows?
>
> 1) Should plots be allowed to be added to multiple windows?

You mean as the file system equivalent to a logical link.

I think 'no'. Drag & drop can copy or move, not link.

For the reasons you stated.

cbn


> if yes:
>
> 2) Should the coordinates and size (aspect - relative to the parent) be
> forced to match in each window?  This gets confusing IMHO.
>
> 3) What happens if the user zooms or scrolls in one window, should it match
> the other?
>
> I think, personally, #2 and #3 indicate that #1 should not be taken
> lightly. It could make it very confusing to work with Kst.  Now to
> circumvent the can of worms that #2 opens up, it is possible to have what I
> am calling a "plot representation", which is an internal representation of
> the plot which contains the actual plot but not the view of the plot.  So,
> for instance, we have Kst2DPlot and Kst2DPlotRep.  If you add plot "abc" to
> main window 1, it would create a new Kst2DPlot which has a pointer to the
> Kst2DPlotRep "abc". Likewise it would happen with main window 2.  Those
> Kst2DPlot would contain a double buffer of the plot and ask the plot rep to
> draw into it if necessary, thus preserving position and scaling but sharing
> zoom, scroll, and other properties.
>
> An unfortunate consequence of #3 is that if all instances of a plot match
> (within the same window and across windows), then we have to propagate an
> update to each view object associated with the rep when there is an
> interactive change.
>
> Therefore these decisions are not to be taken lightly, but I think they
> should be driven by user demand.  They're not something one would hope to
> change after this is implemented, so we should make a final decision now. 
> The view architecture is almost ready to replace the current view so it's
> time to start porting KstPlot.