I haven't formally introduced myself to the group yet, and I'm probably not fully caught up, but why not a zip file (or tarball) containing each dataset as a distinct .csv file along with a manifest file indexing the contents? That's a format that has become fairly common across the industry, and is easily machine unpackable and readable. Thanks, Chris Metcalf Director of Platform chris.metcalf@socrata.comhttp://www.socrata.com On Fri, Feb 21, 2014 at 5:48 AM, Craig Russell <craig@craig-russell.co.uk>wrote: > It's simple, yes, but is it the most appropriate for all contexts? I think > the question of how best to represent multiple data sets (and metadata) in > a single file is worth thinking about. > > On 20 February 2014 17:16, Alf Eaton <eaton.alf@gmail.com> wrote: > >> On 20 February 2014 10:39, Craig Russell <craig@craig-russell.co.uk> >> wrote: >> >> > The downloaded CSV file may include multiple data sets, which are >> separated by a couple of line breaks (example attached). >> >> > There is, at present, no clear machine readable way of differentiating >> these two data sets within a single file. >> >> I like the simplicity of using 2 (or more) line breaks to imply a >> separation between tables (as well to imply the separation between >> descriptive header text and the actual table). >> >> Alf >> > > > > -- > Craig Russell > e: craig@craig-russell.co.uk > w: craig-russell.co.uk > t: @craig552uk >
Received on Saturday, 22 February 2014 22:01:42 UTC