Comments on Model for Tabular Data

Hi All

Danbri asked me to share some thoughts on this spec, from a conversation we
were having elsewhere.  He said the group was interested in comments and
feedback.

Just following on from the discussion on the TAG mailing list.  In the
SoLiD specification [1] we are currently using meta data to associate more
data with resources on the Linked Data Platform.

Having gone through rel="meta" we have now decided to align with this
groups work and use rel="describedBy" for general meta data associated with
files.

We also use rel="acl" for access control lists, but that's not (yet)
registered with IANA, perhaps we will use a full URI.

In general, we will always try and supply the correct header to point
clients to the meta data.  Typically it is of the form .meta or .acl
(though previously we had been using ,meta)

I personally support the points made by David Booth which suggest that
using the .well-known pattern is more appropriate for *per site* wide
configurations, rather than, for *per directory* or *per resource* type
data.

My own interest in tabular data is limited, however one use case I have
looked at in the past is for payments where systems such as paypal would
use a csv for bulk upload of payments.

In our work we are seeing more and more the use of the web, not only as a
cross origin data space, but also as a cross directory, data space.  In
systems where you dont have access to .well-known or are limited to a
directory (dropbox / gdrive / sky drive / box.net perhaps?) I can see in
general a probe sequence starting close to the file's directory making
sense.

Hope that helps!

[1] https://linkeddata.github.io/SoLiD/