I’m finally exploring the new geospatial nodes, with a mixture of excitement and caution.
It’s very exciting to finally have some serious geospatial capability in KNIME, but it’s also apparent that these new nodes are still developing, and in many ways are not production-ready.
Many suggestions for improvement have already been discussed in other threads. My personal wishlist is:
- Integration with KNIME’s file handling framework
- The ability to write visual outputs as images (unless I’m missing something, the visualisation nodes have no outputs at all)
- The ability to select features in a visualisation and have them filter other visualisation elements, or at least save a ‘row selected’ flag akin to the Table Editor.
- The ability to input new data, such as by placing pins or drawing a polygon.
Until the file handling framework is upgraded, it’s hard to see how these nodes can be deployed in a productionised environment. Similarly, if all of the visualisation outputs are purely interactive, with no graphical or selection output, there is only so much we can do with these nodes. (I’m not even sure yet if the outputs are supposed to work in the WebPortal, as I have not succeeded in achieving this yet.)
I gather that these limitations stem from the underlying python integration framework, which I understand is also in its infancy. But it would be really good to have some idea of what improvements we can expect to see over the coming releases. Has a roadmap of this kind been made available, or could it be?
