Process for the status of RDF/SPARQL tests

[afs]

This issue comes from:

Originally posted by @pchampin in #137

I did some digging (only on the RDF part of the test suite, not SPARQL).

Nb of entries attached to a manifest that have a status different from "approved":

-------------------------------------------------------------------------
| g                                           | a                  | nb |
=========================================================================
| <rdf/rdf11/rdf-mt/manifest.ttl>             | rdft:NotClassified | 1  |
| <rdf/rdf11/rdf-n-triples/manifest.ttl>      | rdft:Proposed      | 20 |
| <rdf/rdf11/rdf-trig/manifest.ttl>           | rdft:Proposed      | 4  |
| <rdf/rdf11/rdf-turtle/manifest.ttl>         | rdft:Proposed      | 8  |
| <rdf/rdf12/rdf-n-quads/c14n/manifest.ttl>   | rdft:Proposed      | 37 |
| <rdf/rdf12/rdf-n-triples/c14n/manifest.ttl> | rdft:Proposed      | 37 |
| <rdf/rdf12/rdf-semantics/manifest.ttl>      | rdft:NotClassified | 99 |
-------------------------------------------------------------------------

Nb of entries attached to a manifest that have no approval status

------------------------------------------------------
| g                                             | nb |
======================================================
| <rdf/rdf11/rdf-mt/manifest.ttl>               | 8  |
| <rdf/rdf11/rdf-n-quads/manifest.ttl>          | 2  |
| <rdf/rdf11/rdf-n-triples/manifest.ttl>        | 48 |
| <rdf/rdf11/rdf-trig/manifest.ttl>             | 2  |
| <rdf/rdf11/rdf-turtle/manifest.ttl>           | 2  |
| <rdf/rdf12/rdf-n-quads/c14n/manifest.ttl>     | 4  |
| <rdf/rdf12/rdf-n-quads/syntax/manifest.ttl>   | 27 |
| <rdf/rdf12/rdf-n-triples/c14n/manifest.ttl>   | 4  |
| <rdf/rdf12/rdf-n-triples/syntax/manifest.ttl> | 29 |
| <rdf/rdf12/rdf-trig/eval/manifest.ttl>        | 25 |
| <rdf/rdf12/rdf-trig/syntax/manifest.ttl>      | 35 |
| <rdf/rdf12/rdf-turtle/eval/manifest.ttl>      | 29 |
| <rdf/rdf12/rdf-turtle/syntax/manifest.ttl>    | 67 |
| <rdf/rdf12/rdf-xml/eval/manifest.ttl>         | 30 |
------------------------------------------------------

I think we should decide on a consistent policy and stick to it.

My understanding is that all entries attached to a manifest are implicitly approved. So either we mark them all as approved (updating or adding the rdft:approval property), or we drop the rdft:approval completely for all these entries. I prefer the second option, following the DRY principle.

There are also a few floating entries in the manifest files. Should we keep them??

[rubensworks]

As the mode now is continuous improvement by the wider community, marking the WG acceptance by git hash, permanent tag or branch, or zip file is more appropriate than expecting the user to filter the current state.

This (by @afs in #137 (comment)) sounds like a good approach to me.
It lowers the overhead of managing approvals per test, while still allowing the WG to put a "stamp" on a full test suite.