Is there a preferred way to include schema definitions with a dat?
How might one handle datasets with multiple related tables?
Do you see dat as similar to Qu in its longer term ambitions? Complementary?
See the Qu roadmap.
On multi-table support, see @maxogden’s response to John Resig’s query.
On dat schemas, see these comments on the dat freenode channel …
… and this discussion on package file conventions for datasets.
@jbenet notes that he’s working with @maxogden with the aim of making dat/datadex interoperable. See also @maxogden’s comments at the very end of this article.
@maxogden has stated his preference for keeping metadata and data in separate files (i.e., having a distinct file with metadata, rather than having a header with metadata included with the datafile).
@mafintosh has discussed a schema.json file describing the global encoding/decoding format for each row of data in dat.
Note also the proposals for a dat schema api in #171 and #191.
See, finally, this proposed vision for a dat registry.