We should have features, points and polygons tables

[IanHopkinson]

We should potentially put points in a separate table called points and rename the swdata table features with a feature_index in each table as a common key. This would make the handling of the different datatypes equivalent but have the downside of making data download more complex - someone just interested in points would have to download two separate files.

[drj11]

There are plenty of datasets with 0 points per feature (eg, a polygon based country dataset). Are there any datasets with more than one point per feature? If no, or very few, I would suggest keeping points and features together (in a renamed feature table). More practical, but less elegant.

[IanHopkinson]

OK - renamed swdata to feature