Re: [udig-devel] Render Problem

[Jesse Eichar <jeichar@xxxxxxxxxxxxxxx>]

On Wed, 2005-03-16 at 01:53 +0000, Tony Kennedy wrote:
> The platforms are windows and Linux and the shape file is on the local
> drive. I was mystified why we had not seen this before but on
> reflection it is the introduction of the style feature and easy access
> to text labels that gave rise to the use of the node layer (people can
> now see the distance between road nodes and compare vehicle miles
> traveled). The problem exists with information nodes as well. The
> files display correctly on our arc box and OSGB viewer, not conclusive
> I know. But the cancel operation should have worked no mater what. The
> render operation completes successfully and other operations can take
> place indicating the file set is correct I can send the file set but
> it is 95m unzipped which may fall foul of your email system let me
> know if you require it.

I believe that I've fixed one bug with the cancel.  However there are
still some issues to be resolved.  The following is a synopsis of why
cancel does not always work correctly.

The poor cancel button does not have the power to interrupt io.  The
execution is something like the following (Geotools people correct me if
I'm wrong):

-LiteRenderer begins renderering pipeline.
-FeatureReader is opened.
-each feature is processed sequentially from reader.  After each feature
is processed (or even while the features are being processed) the stop
flag is checked.
-when the stop flag is set the reader's close method is called.  
-Rendering stops after the close method finished.

For a while the WFS data store would not stop its processing of its
incoming data when the stop method was called so the renderer blocked on
the close method.  I think that WFS, shapefile and postgis work fairly
well. 


Jesse