[udig-devel] Some plotting issues and questions

[Jesse Eichar]

>>
>> Further to my previous email I am now trying to plot data using  
>> only WFS layers and a few issues / feature requests / bugs have  
>> come to light, hopefully the images attached will help illustrate  
>> and someone can point me in the right direction...
>>
>> Snapping to points on any layer
>> this is great but is it possible to show all the nodes on the map  
>> so I can see what I'm snapping to before I click the mouse  
>> button?  Image 1 (attached) shows a curved line in my base data  
>> that I want to snap to.  (The circle is the cursor, no idea why  
>> the crosshair didn't get captured in the screen dump?!?!) Image 2  
>> shows the same line highlighted for editing so that I can see  
>> where the nodes are.  I would like to see this without having to  
>> select it for editing, is this possible?  Would it also be  
>> possible to show which node will be snapped before the mouse  
>> button is clicked?
>
> Doing more than just providing a circle is extremely difficult  
> simply because the data must be loaded from the data source which  
> is a slow process.  Loading the entire layer is prohibitive because  
> it is difficult to know how much information you might get when you  
> try to do the load.  We try to keep as little in memory as  
> possible.  I have found that even with the current editing  
> framework it requires a great deal of memory.  Loading more would  
> cause serious problems.  So I'm afraid that it is kind of up to the  
> user to try to figure it out.  What can be done is the line can be  
> editing after the fact by moving the vertex again.
>
> Seems to be some confusion on this one, what I'm asking for is the  
> ability to show the nodes on a layer.  Only the loaded part of the  
> layer and surely the nodes must have been loaded into memory in  
> order to render it in the first place?  It would be nice if when  
> snapping the closest node to the cursor could be highlighted but  
> that's less essential than showing the nodes in the first place.   
> Maybe this is already possible with styles?
>

Yes of course the features are rendered but sometimes the view can  
contain a massive amount of information.  Consider a shapefile  
millions of features and you are zoomed out to the whole world.  You  
probably don't want all of that in memory.  So how do you know when  
to keep things in memory.  One thing we've considered is to provide  
the option to 1.  Load a layer into memory and 2.  be able to load a  
portion of a layer into memory.  If the user activates one of those  
mode then this would be possible.  (This isn't implemented yet).

[...]



> when can we expect a release based on Geo-Tools 2.4 as per my  
> previous email?  I'm hoping that will mean I can load the base  
> table data direct from Oracle instead of through WFS which would  
> presumably make it faster?

It will be a little while as there were some significant API change  
from Geotools 2.2. and 2.4.  We've got it running but there are still  
bugs and some of the styling work is still requires some repair.

Jesse
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.refractions.net/pipermail/udig-devel/attachments/20070720/eff1af27/attachment.html