Hi, this is to explain what I have in place in order to handle what we're
calling "pointer datasets" in CDAWeb.  These are datasets where we have no
actual data files in the system, but where we've established a dataset so
that we can point to project/PI provided links.  Examples of this are the
TIMED, DMSP and GOES12(EPS) datasets.

What's currently required in order to "install" a pointer dataset is the
following:
1) install a master cdf w/ cdawput -M datasetname.cdf
2) place a corresponding dataset data cdf file in the
   /home/cdaweb/data/pointer_files subdirectory, where the file is a copy
   of the the master w/ a single variable called EPOCH which has two
   records a start and end record that must span a time of exactly 1 hour, e.g.
   minimum: 22-Jan-2002 00:00:00, maximum: 22-Jan-2002 01:00:00
   This one hour is what the eval2.cgi script keys off of to know not to
   show the dataset time range in the CDAWeb Explorer page, and it also
   won't be included on our inventory plots.  Not elegant, but its the
   best I could come up with at the time.
3) check out CampaignDefinitions file w/ cdawget -l CampaignDefinitions
   edit the file, adding the datset name in the appropriate "catalog"
   check it back in w/ cdawput CampaignDefinitions
   make sure to specify what dataset you added and where when prompted
   for a comment.

TJK 6/23/2003