Hello all,
I think it is a great idea to allow multiple conventions in netCDF
files. As far as I know, the talk I gave at Unidata several years ago
(http://docs.google.com/viewer?a=v&pid=sites&srcid=ZGVmYXVsdGRvbWFpbnxnYWxlb250ZWFtfGd4OjI2MTdh&pli=1)
was the first mention of the Metadata_Conventions tag. Obviously, if we
can put multiple conventions in the Convention tag that can work as well.
That talk also included the idea of a link back to the complete
metadata. We suggested the tag name metadataConvention_Link (FGDC_Link
in the example I showed, must be old!). I think that this link is a
critical part of the equation that should also be standardized.
As far as conflicts between different conventions, I don't think this is
a problem as long as the tag names are different. Also, the amount of
metadata is generally small, so having redundancy is not really a problem.
Ted
Ethan Davis wrote:
Hi Nathan, all,
Nathan Potter wrote:
Following a fairly long discussion about identifying metadata
conventions (1st message here:
http://www.unidata.ucar.edu/mailing_lists/archives/thredds/2010/msg00008.html
) we arrived at a road block due to differences in the way three
documents describe the mechanism through which one would identify
the list of metadata conventions used in the data:
This discussion has been brought up on the cf-metadata list. Here's a
good summary of the current CF activity on this front:
On the cf-metadata list, Jonathan Gregory wrote
Both of your discussions are somewhat related to CF trac ticket 27
https://cf-pcmdi.llnl.gov/trac/ticket/27
In the latter part of that, there is generally agreement that we should allow
other conventions to provide attributes which are labelled with a prefix, as
Heiko suggests for including a discovery metadata convention. There seems to be
no problem with that, so long as the other convention is not providing the same
kind of metadata as CF, so there will not be any contradiction.
There has been other discussion recently on this email list concerning the
issue of naming other conventions in the Conventions attribute. I don't think
the original intention was to exclude that possibility. It's just not
recognised in the CF standard and it should be. However, no-one's had time
propose an amendment. Personally, I think it's OK so long as the extra
conventions accept all of CF, and just add more conventions which do not
conflict. If they overlap, this has to be thought about carefully, and in that
case I would say that the CF standard would have to be amended to describe how
the overlaps should be resolved.
I'll spend some time working on CF trac ticket 27 and hopefully get that
moving forward. With all this recent interest, I'm guessing we'll have
some movement on rewording the "Conventions" attribute section as well.
I will also take a look at the "NetCDF Attribute Convention for Dataset
Discovery" document with an eye towards modifying, backwards
compatibility, and any changes to CF.
--
Ethan
Nathan Potter wrote:
Greetings,
Following a fairly long discussion about identifying metadata
conventions (1st message
here:
http://www.unidata.ucar.edu/mailing_lists/archives/thredds/2010/msg00008.html
) we arrived at a road block due to differences in the way three
documents describe the mechanism through which one would identify the
list of metadata conventions used in the data:
NetCDF Conventions:
http://www.unidata.ucar.edu/software/netcdf/conventions.html
CF-1.4:
http://cf-pcmdi.llnl.gov/documents/cf-conventions/1.4/cf-conventions.html#identification-of-conventions
NetCDF Dataset Discovery:
http://www.unidata.ucar.edu/software/netcdf-java/formats/DataDiscoveryAttConvention.html
The problem:
The NetCDF COnventions page identifies an attribute called
*Conventions* to be used to provide a list (space or comma separated) of
one or more metadata conventions used in the file. In the CF-1.4 page,
section 2.6.1 Identification of Conventions provides a description of
the use of the *Conventions* attribute that can easily be (mistakenly)
interpreted to mean that the *Conventions* attribute may not contain a
list. In the NetCDF Dataset Discovery page the *Conventions *attribute
is replaced by a *Metadata_Conventions *attribute that appears to be
synonymous with the previously defined *Conventions* attribute.
A solution:
1) Clarify the CF-1.4 page so that it is explicitly clear that a space
or comma separated list of conventions is allowed.
2) Amend the NetCDF Dataset Discovery draft so that it
uses* Conventions. *Either by replacing *Metadata_Conventions *with
*Conventions*, or by explicitly stating that *Metadata_Conventions* and
*Conventions* are synonyms (semantically identical)
Can you guys get together and make that happen? Am I asking a lot?
Nathan
_______________________________________________
thredds mailing list
thredds@xxxxxxxxxxxxxxxx
For list information or to unsubscribe, visit: http://www.unidata.ucar.edu/mailing_lists/
--
==== Ted Habermann ===========================
Enterprise Data Systems Group Leader
NOAA, National Geophysical Data Center
V: 303.497.6472 F: 303.497.6513
"If you want to go quickly, go alone.
If you want to go far, go together"
Old Proverb
==== Ted.Habermann@xxxxxxxx ==================
