[OSGeo-Discuss] OSGeo Journal Volume 11 (2011 Annual Report) Published

[Adrian Custer]

Hello,

Unfortunately the piece in the OSGeo Journal, vol 11, on 'Simple 
Features' is promulgating a common misconception.

It turns out that 'Simple Features' are *not* constrained to have a flat 
table of attributes. 'Simple Features' are only restricted to have a 
single geometry which must be 'simple.' The attributes may be 
arbitrarily nested and may be in different namespaces. Also the 'simple' 
geometry, which traditionally included only elements in 2D with linearly 
interpolated arcs (points, lines, polygons) are now being expanded to 
more forms and possibly more dimensions. So it turns out that 'Simple 
Features,' as discussed in the standards group, are not so simple.

We do need a term to talk about the kinds of features discussed in the 
article. The best I have discovered is 'Flat Feature' which implies:
     1) no explicit discussion of temporality
     2) a single spatial definition, which must be 2D and only use 
linear interpolation
     3) have a 'flat' table of other attributes, that is that attributes 
are not nested in each other.
The 'flat' in the name refers both to the geometry and to the form of 
the attributes, which could form a single row in a table.

cheers,
   ~adrian