Ben,
Good one about TCP!
The only downside I see is the resources it takes to move it through
the process. I think the time *probably* would be well spent; I say
'probably' only because I think it will take a LOT of time to (a)
create a description of the specification that is suitably clear and
comprehensive, and (b) get enough time working on it to get through
the process.
Well, there's a third potential time issue: I for one would want to
see a few changes in CF before I would say it should be approved via
OGC. And there may be others with their own concerns. So first we'd
have to discuss whether OGC should rubber stamp the existing standard,
given its huge community and 20TB of existing data, and worry about
improvements later; or whether there is a 'minimum bar' of
interoperability that has to be supported for any OGC standard.
(Presumably there are some criteria, or ESRI binary shapefiles would
have been accepted. I don't know the history of that, though. Are
there a set of criteria that get applied to every standard in OGC?)
To desensitize any discussion on that point, let me cite an
archetypical example, Recently I saw a complaint about another
community protocol that has no embedded metadata and is very hard to
parse. The protocol has been in use worldwide for a few decades and
may have thousands of users (certainly hundreds), transmitting data
real time 24x7 all that time. So it's been very successful in that
sense. The question is, just because it has been shown to work, and
is widely adopted, is that enough to be an interoperability standard?
Or should a standards body say "These are minimum criteria that any
standard must fulfill"? If the latter, I am curious, what are those
criteria?
To avoid sending us totally off-topic here, let me return to my
conclusion that I think it would be healthy, for both the science
community and the CF-netCDF community, if CF-netCDF went through some
standards process like OGC. But it might be painful too.
John
On Aug 24, 2009, at 3:10 PM, Ben Domenico wrote:
Hi all,
These are really valuable discussions. In my mind they are just as
important as the formal standards that result from that part of the
process. In the various OGC working groups where I've been active ,
I think we all have a much better understanding of the other
subgroups needs and their approaches to satisfying those needs. I
certainly count myself among those who have received one heck of an
education over the last few years.
In the current discussion though, one point I still don't grasp is
what is to be gained by NOT specifying CF-netCDF as A standard for
binary encoding. Not THE standard necessarily, but one possible
formal standard option. It's as if people think that CF-netCDF is
more likely to be replaced by a newly minted standard if CF-netCDF
is not declared a standard. Those of us who've been at this long
enough to remember the declaration of the ISO OSI transport layer in
the late 70s realize that the non-standard TCP still has a modest
following in many communities.
In the case at hand, I'm really convinced that it's a good idea to
build on proven technologies while AT THE SAME TIME working on
specifications (e.g., SOS, WFS, WCS, SWE common, ...) that may be
more comprehensive, fill gaps and address shortcomings of the
existing approaches -- approaches that have been shown to work, but
may not be all things to all people. As we proceed, it's essential
to keep this valuable dialog going so the individual components have
a chance of fitting together in some sort of coherent whole in the
end.
-- Ben
On Mon, Aug 24, 2009 at 3:26 PM, John Graybeal <graybeal@xxxxxxxxxxxxxxxxxx
> wrote:
On Aug 24, 2009, at 10:42 AM, Steve Hankin wrote:
NetCDF (& associated tooling) is arguably emerging as the
definitive standard for interchange of 3-dimensional, time-
dependent fluid earth system datasets.
For the members of the NetCDF community who favor this argument, may
I point out there are other communities that say similar things
about their solutions? And I'm not referring to OGC, which to my
knowledge has never pitched SWE (or anything else) as a straight
replacement for NetCDF, notwithstanding Alex's claims for SWE's
representational capabilities. I mean, it's not like apples and
zebras, but the two seem really different to me.
I like NetCDF for a lot of things, including many-dimensional and
time-dependent data representations.
But terms like "definitive standard" carry their own hyperbolic
weight, especially in a world of multiple standard bodies and many
different kinds of system requirements.
So it seems to me there will not be *a* winner, either in this
argument or in the earth science data management community's choice
of technologies. Thus, I'm much more interested in understanding
the characteristics of each, so as to use them well and maybe even
improve them. (Hmm, I suppose that would explain my project
affiliation....)
John
---------------
John Graybeal
Marine Metadata Interoperability Project: http://marinemetadata.org
graybeal@xxxxxxxxxxxxxxxxxx
---------------
John Graybeal
Marine Metadata Interoperability Project: http://marinemetadata.org
graybeal@xxxxxxxxxxxxxxxxxx
