EARL 1.0 Proposals and Choices

[Non-exhaustive.]

Proposals

* Remove earl:excludes [and possibly earl:suite] or move it to a
utility namespace
Rationale: experiments with EARL 0.95 have proven that whilst reducing
exclusions (even with additional suites) is remarkably easy to
accomplish with CWM, applying them is quite difficult, unless you know
the precise number of exclusions (applying one exclusion is easy).
Custom tools may be able to work around this. In any case, the whole
"exclusion" and "suite" concepts are simply shorthands for explicitly
enumerating lots of test point IDs.

* Use the new resource/representation/documentation suggested in my
message of
http://lists.w3.org/Archives/Public/w3c-wai-er-ig/2001Dec/0038 [1]
Rationale: more consistent with Web architecture, allows for
XPointers,

* Add earl:EquivalenceRelationship
Rationale: will let people state their own equivalence relationships
if need be. This is mainly processor specific, and builds upon the
"earl:sameContentAs" property in EARL 0.95. This will allow for
hashing properties to be made, etc. An equivalence relationship is one
which is reflexive, transitive, and symmetric. Note that only
transitivity is covered by DAML, although the others may be covered by
the upcoming OWL.

* Move confidence levels and other less-used features of EARL to a
utility namespace
Rationale: We could sugest that canonical processors needn't bother
with them in that case.

Choices & Issues

* Arbitrary naming Automaton vs. Machine, Human vs. Person

* Name for the testSubject variant for tools. Suggestion:
earl:documentation

* Use outside properties? foaf:mbox vs. earl:email, rdfs:comment vs.
earl:note

* We need a "loose note" and a "normative definition" set of
properties. I had a (very) drief discussion with DanC about this, and
he seemed to suggest just putting them up on the Web somewhere. I've
used a little bit of infomesh.net space for it for now, but I'm not
all that comfortable with it - I like namespaces to be in persistently
maintained space. Well, I'd like that for EARL. That's my grudge
against not using properties like foaf:mbox... the old issue of
namespace handling seems less important now. It's would be good for
the Semantic Web if we just used any old stuff that we found on the
Web (that suits our requirements). As Seth said on swag-dev, "we are
fighting the Tower of Babel".

* Datatypes. RDF Core still aren't decided on this. If EARL 1.0 gets
released before they decide, we'll just have to go with the favourite,
and perhaps change it later.

Cheers,

[1] [22:51] <sbp> .google 00b301c1836c$d0857b80$ceb80150@localhost
[22:51] <xena> 00b301c1836c$d0857b80$ceb80150@localhost:
http://lists.w3.org/Archives/Public/w3c-wai-er-ig/2001Dec/0038.html
[22:51] <sbp> ah, mid: to http: resolution

--
Kindest Regards,
Sean B. Palmer
@prefix : <http://purl.org/net/swn#> .
:Sean :homepage <http://purl.org/net/sbp/> .

Received on Tuesday, 29 January 2002 18:05:29 UTC