Just one note, quickly added:
Wright, Bruce wrote:
1. Coverages and features are different...WFS and WCS evolved as two
distinct services to meet different requirements for accessing data and
metadata.
this is past and presence, IMHO.
2. A coverage is a feature...features and coverages are different
'cross-sections' through the information - Simon Cox presents this
nicely by considering the information as tabular, with a row represents
a feature (a series of individual property values) and a column
representing a coverage (different values of the same property) - and
the WFS and WCS should be harmonised.
hm, that seems like adding a third, radically new concept to unify the
two others.
Why not simply say "a coverage is a feature which enjoys special
treatment, as laid down in the WCS".
3. A feature is a coverage...coverages are already effectively being
encoded in GML for some WFS requests that need to return the variation
of a set of parameters over space/time (normally small data volumes);
again, this suggests that the WFS and WCS should be harmonised.
well, I have nothing against GML as one _additional_ data format (and a
proof that something is possible). All the mapping people I have talked
to, however, want to first webify their vector material and then their
rasters - current practice, alas, has made WFS the first-born son ;-)
4. Coverage is a property of a feature... WCS is a convenience
interface, which should eventually replaced by an enhanced WFS, which
adds a GetCoverage request (or an OPeNDAP request!)
oops, that sounds complex - just a property (aka attribute) of a feature?
We might adopt #2 and come to the same conclusion.
Personally, I think these are all true to some extent (not sure 3. above
is a good thing though!). However, which viewpoint you take determines
how you develop and implement these web services going forward (e.g. my
explicit 'conclusion' on 4. above!).
I very much agree, this is determined by history and current practice.
Nice discussion!
nite,
Peter
