previews, tests

Followup on the note on previews and such; what about a conventional
way to represent this? E.g. something like:



(Doe, 1999a, 1999b; Smith, 2000)



  • Doe, J. (1999) ...
  • ——— (1999)...
  • Smith, S. (2000)...

The csl-bibliography div could just be embedded in the Atom entry for
a style. A tool like Zotero could then allow users to browse for the
style they want to use (and “activate”).

In related news, Tim Bray had something yesterday about the Atom
Publishing Protocol (APP) and related MS alternative that reminds me
of previous conversations here:

“Test suites matter way more than specs, in the big picture.”

For this work, I guess that really means a set of example data that
propertly tests a) whether an implementation supports the features in
the spec correctly, and b) whether a CSL style is correctly
constructed to reflect the expectation of the style.

It seems to me a minimal set of examples goes pretty far to test:

  • author-date sorting/grouping and disambiguation
  • the type-based inheritance (knowing to treat an album like a book
    and so forth)
  • ibid
  • first/subsequent
  • missing author substitution (use editor, else pub title)

I started to do this in RDF (N3) awhile back, but it’d obviously need