Hi Ben:
If the netcdf encoding is just an extension (rather than at the core
of WCS), and if it is not supported by the larger WCS or OGC
community (ie. if most WCS clients or servers do not support it),
then it is not serving the purpose of providing interoperability
between communities, which would seem to be the purpose of this effort.
The GALEON effort has been very strong, and a lot of great work has
been done. But is this viewpoint supported by the larger WCS or OGC
community? Recent decisions about WCS would seem to suggest not. I
would return to my email of last week. Where in OGC land is there
something that is close in viewpoint to that of the metocean
community? IMO it is CSML - the mapping between CSML Feature Types
and Datatypes in the proposed CDM is very close. Rather than try to
force a harmonization on the WCS level, where the larger WCS community
does not appear to be interested, why not work for closer
harmonization with CSML, with gateways and translators, and since the
CSML folk are heavily involved in OGC, use whatever OGC transport
layers/ service requests they develop for CSML as the ones that will
provide the bridge.
Sort of half-formed ideas (Steve is always much more articulate than
I), but I think you get the idea.
-roy
On Sep 23, 2008, at 3:19 PM, Ben Domenico wrote:
Hi Steve,
First, thanks to you and Jon for your thoughtful emails. It will
take me a while to digest them more fully and come up with a follow
up to the discussion -- especially the OPeNDAP suggestions.
But there is one point that's a bit puzzling to me. I agree with
the view of "WCS as an OGC-accepted vehicle for interoperability
*between* our community and others." But I don't see how it follows
that "It is not clear that embedding netCDF into WCS any longer
holds promise of contributing to that goal." To the contrary, my
own sense is that great progress has been made toward that objective
and Stefano's draft encoding specification gets us over the last
remaining major hurdle.
What have I missed here?
-- Ben
On Tue, Sep 23, 2008 at 12:02 PM, Steve Hankin <Steven.C.Hankin@xxxxxxxx
> wrote:
All,
Jon has done an excellent job of expressing the fundamental concerns
that I also see. I ask your indulgence at for tossing yet-more
radical ideas into the ring.
It remains imperative that our "fluid earth science" community
continue to explore all useful angles of connection with the OGC
community. The contributions that GALEON has already made in this
regard are enormous, and from all that I have observed are very
broadly recognized and appreciated. For myself, I cannot thank you
enough. Finding this much time and energy to devote to community
standards while also meeting "day job" demands is extraordinary.
I would like to pose this question: is GALEON at a point where it
should take stock, re-evaluate goals, and possibly make a
significant course change? That is a very hard thing for an
individual to do, and a much harder thing for an organization to
do. With the WCS 1.2 "core plus extensions" approach, we have
effectively seen the broader OGC community reject involvement in the
level of complexity that our community requires in order to conduct
its internal business. As Jon says, arguably the central goal of
GALEON was to explore WCS as an OGC-accepted vehicle for
interoperability *between* our community and others. It is not
clear that embedding netCDF into WCS any longer holds promise of
contributing to that goal.
If this interpretation of recent decisions is correct, then how
should we adapt? What approaches will best meet our community's
needs? With the combination of netCDF, CF, and OPeNDAP we have a
powerful and effective vehicle for interoperability that has already
gained quite broad acceptance within much of our community. The use
of OGC-accepted solutions is, however, mandated for many of our
partners, and that has created a schism. A true fusion of OPeNDAP
and OGC could mend that schism and lead to a huge interoperability
success story for our community. How can we best blend the virtues
of netCDF/CF/OPeNDAP with the OGC process? Is there work that
GALEON could pursue to "bless" OPeNDAP transport under an OGC/WCS
banner? Can we envision, for example, a community extension to WCS
that returns an unconstrained OPeNDAP URL as the payload of a WCS
data request? This would essentially provide an OGC-accepted segue
from WCS into OPeNDAP. With relatively minor effort at relinking,
any application that can read netCDF can read OPeNDAP, so this is
not far from what GALEON is currently exploring using netCDF-CF
files as a WCS payload.
And how to approach interoperability with users outside of our own
community? Jon's suggestion of "the WCS specification considerably
simplified" seems very sensible. And this approach would dovetail
well with the recent work on server-side regridding through
OPeNDAP. It is easy to envision OPeNDAP servers as gateways --
providing simple lat-long versions of our gridded data that are then
formatted as geoTIFF files and delivered through the simplest WCS
syntax.
2 cents (or maybe 4) - Steve
