Question about ogrlayerinfo and wfslayerinfo (layer inheritance needed)

[Daniel Morissette]

Tamas Szekeres wrote:
> 
> I can see the problem by now. WFS layer extends the functionality of the OGR
> layer and each of them should maintain separate layer information.
> 

Correct.

> In fact the problem could be solved if we could establish a similar concept
> like inheritance in the OO design. WFS layer should be inherited from the
> OGR layer by adding the WFS specific layer information.
> 

Yup, having the WFS layer struct as an extension of the OGR layer struct 
should do. There's a few tricks we can do with C structs to emulate that 
kind of behavior.

> I am thinking of a solution targeting this problem that could allow a layer
> provider to extend the functionality of another provider. The recently added
> vtable architecture makes a good foundation to achieve this. It could also
> bring in great additions of the functionality in other areas.
> For example:
> 
> - we could implement a provider independent feature caching system to
> increase the rendering speed of the applications that can maintain the layer
> object state between the drawings. By reducing the access to the providers
> could decrease the thread interaction of the application as well.
> - we could implement various provider independent data filters.
> - by creating a "layer extender" we could increase the rendering area at the
> provider side to allow drawing of the annotations falling onto the map
> borders (see the related kamap label splitting issue)
> - and many more...
> 

Hummm... this is definitely more than we need for this specific issue 
which is simply that WFS layers use an OGR connection to read the GML.

Daniel
-- 
Daniel Morissette
http://www.mapgears.com/