[eml-dev] Recommendations for reporting file access errors to software agents

Wade Sheldon sheldon at uga.edu
Tue Apr 12 16:55:51 PDT 2005


Matt and all,

What type of error message (content and formatting) do you think should be
returned to a software agent that tries to retrieve an inaccessible ASCII data
table referenced in an EML document? For instance, if a particular version of a
data set has been taken offline due to the discovery of major errors, how should 
the file access error be communicated to Kepler or some other Metacat/EML-savvy 
software agent (i.e. when they try to retrieve the old file based on a legacy 
EML document that hasn't been re-synced yet)?

The reason I'm asking is that I've been revising the GCE data distribution web
application to support stored cookies and more gracefully handle requests for
outdated or offline files, and I'd like to provide optimal support for both
interactive web clients and agents. I am currently using http and session
characteristics to distinguish these client types, and bouncing web clients to a
"contact the IM for assistance" web form on errors and returning a simple text 
message to agents, e.g. "Error: PLT-GCED-0409_1_2.TXT was not found on the 
GCE-LTER server (contact the GCE Information Manager <gcelter at uga.edu> for 
assistance)". If there is a common structured error message format (xml or text) 
supported by Kepler or other SEEK tools, though, I will use that instead.

Here are examples of what I'm doing now for invalid versions of data set tables 
(invalid data set accessions or missing files are handled similarly with 
different messages):

Invalid EML distribution URL (agent client):
http://gce-lter.marsci.uga.edu/lter/asp/db/send_file.asp?accession=PLT-GCED-0409&filename=PLT-GCED-0409_1_1.TXT&name=metacat-user&email=none&affiliation=LNO&notify=0

Invalid EML distribution URL (interactive client, using web session/cookie/login):
http://gce-lter.marsci.uga.edu/lter/asp/db/send_file.asp?accession=PLT-GCED-0409&filename=PLT-GCED-0409_1_2.TXT


One a related note, I've been archiving outdated but otherwise valid data sets
(e.g. those that have been updated to add an attribute or records) so that they
can be retrieved by URLs in legacy EML docs. However, not communicating the fact
that a more recent version of a data table is available seems to be a disservice
to users. Do you think it would be preferable to return an error or message if a 
more recent version of a data set exists, or just keep returning the old version 
as requested (if not taken offline) until the metadata is re-synced?

We are currently finalizing the "EML Best Practices for LTER Sites" document for 
general distribution, so if there is a consensus on these issues I'll add 
recommendations to the eml/dataset/dataTable/physical/distribution/online/url 
section.

Thanks.

Wade Sheldon
GCE-LTER Information Manager


____________________________________

Wade M. Sheldon
GCE-LTER Information Manager/SIMO Database Administrator
School of Marine Programs
University of Georgia
Athens, GA 30602-3636
Email: sheldon at uga.edu
WWW: http://gce-lter.marsci.uga.edu/lter/bios/wsheldon.htm



More information about the Eml-dev mailing list