Re: [udig-devel] Tiled Rendering and Mapgraphics Again

[Emily Gouge <egouge@xxxxxxxxxxxxxxx>]

Hello Jody,

I'm slowly getting back into the udig mapgraphic rendering stuff.  I 
have looked at this and am happy to change the workflow as you have 
suggested.

However I have one question regarding this statement:
"to see the same bugs I am seeing open up a map and hit the "New Page" 
button in the toolbar"

I don't see any 'New Page' button - what am I missing?

Emily


Jody Garnett wrote:
> Okay the problem was that the layer was null; no that does not make
> much sense to me either :-)
>
> However the other way of looking at this is that Renderer has two
> methods (one rendering onto a provided Graphics2D, and one for
> rendering onto a raster held in the context).
>
> So we need to change your workflow a bit:
> - MapGraphicRenderer.render( Graphics2D destination, IProgressMonitor
> monitor ) should not use caching - it is "raw" and draws directly to
> the provided destination
> - MapGraphicRenderer.render( IProgressMonitor monitor ) is called to
> render into the provided context; I think this may be where the code
> for your cache can live? After it sets up the cache image it can use
> the first method to draw into it?
>
> In anycase I am a bit confused on this one; to see the same bugs I am
> seeing open up a map and hit the "New Page" button in the toolbar;
> this will create a new page editor - and it is the MapPrinter box that
> is having all this trouble.
>
> Jody
>
> On Sat, Mar 14, 2009 at 8:58 PM, Jody Garnett <jody.garnett@xxxxxxxxx> wrote:
> > Okay my first useful feedback - as part of the code sprint I am
> > looking into printing in landscape mode.
> >
> > And one of your changes is producing a NPE.
> >
> > // this line is causing trouble when the renderer is used for printing
> > //
> > BlackboardItem cached =
> > (BlackboardItem)getContext().getLayer().getBlackboard().get(BLACKBOARD_IMAGE_KEY);
> >
> > I am going to see what I can be done.
> > Jody
> >
> >
> > On Thu, Mar 5, 2009 at 11:01 AM, Emily Gouge <egouge@xxxxxxxxxxxxxxx> wrote:
> > > Hey Jody,
> > >
> > > What I have done is update the Mapgraphic Renderer to render the entire
> > > viewport onto a single image.  This image is then stored on the layer
> > > blackboard and used when rendering tiles.  When the tile comes along to
> > > render the mapgraphic the image is taken off the blackboard and the part
> > > that matches the tiles bounds are clipped and drawn.
> > >
> > > In order to make this work; when the mapgraphic context is created a
> > > viewport listener is added to the map so when the viewport bounds change the
> > > image cached on the layer's blackboard is cleared.
> > >
> > > The issues with the styling is not that layer.refresh() is not being called
> > > but rather that this doesn't cause my "cached" mapgraphic image on the
> > > blackboard to be cleared.
> > >
> > > A simple solution would be do not cache this image; but I can't imagine that
> > > would be very good performance wise as it would be rendering the image
> > > multiple times.
> > >
> > > I could add another listener that listeners for style events and clears the
> > > cache.
> > >
> > > Or maybe I could store information about the styles with my cached image and
> > > if they have changed then refresh the cached mapgraphic.
> > >
> > > Emily
> > >
> > >
> > > Jody Garnett wrote:
> > > > Hi Emily.
> > > > On Thu, Mar 5, 2009 at 9:48 AM, Emily Gouge <egouge@xxxxxxxxxxxxxxx>
> > > > wrote:
> > > >
> > > > > The ongoing issue of tiled rendering and mapgraphics is causing me issues
> > > > > again so I'm working on a solution.  (The mapgraphics currently don't
> > > > > render
> > > > > properly if using tiled rendering).
> > > >
> > > > Good to hear you making progress on this; when you have a moment I am
> > > > interested to hear what approach your solution is taking.
> > > >
> > > > > I currently have a solution that works most of the time.  The only thing
> > > > > it
> > > > > doesn't do properly right now is refresh when the style is changed.
> > > >
> > > > Interesting. So the style changing would change the layer blackboard which
> > > > would fire an event (or EMF Notification?). From their we need to check to
> > > > see who is listening and make sure they issue a layer.refresh( null )
> > > > event?
> > > >
> > > > > I'm wondering if somebody can review my proposed changes and provide me
> > > > > with some feedback and ideas for dealing with the style change issue I'm
> > > > > having.  If somebody is willing to review the code I'll pass it along to
> > > > > them.
> > > >
> > > > I have a bit of a backlog on uDig items right now; so I cannot in good
> > > > faith
> > > > volunteer at this time (I still owe Anrdrea a style save api).
> > > >
> > > > Jody
> > > >
> > > >
> > > >
> > > > ------------------------------------------------------------------------
> > > >
> > > > _______________________________________________
> > > > User-friendly Desktop Internet GIS (uDig)
> > > > http://udig.refractions.net
> > > > http://lists.refractions.net/mailman/listinfo/udig-devel
> > > _______________________________________________
> > > User-friendly Desktop Internet GIS (uDig)
> > > http://udig.refractions.net
> > > http://lists.refractions.net/mailman/listinfo/udig-devel
> _______________________________________________
> User-friendly Desktop Internet GIS (uDig)
> http://udig.refractions.net
> http://lists.refractions.net/mailman/listinfo/udig-devel