Re: EARL 1.0 Test Schema

"Sean B. Palmer":

>* earl:ccppSetting is now just a strawman property

Whilst I agree with this, what about content negotiation, that does exist
and is quite commonly used, language, gzip very common - but probably
make little change to the accessibility of a report although language
could influence things like line numbers/xpointer etc.   (also should we
be using line numbers rather than xpointer, could we offer both?)

I also still think some property that allows you to check that what
you're getting is what the test was done against, an MD5 hash of the url
would seem appropriate to this.

I'm not sure what the point of Platform and operating system are, what do
they offer, and how would they be used?

>* earl:testSubject is no longer a daml:UniqueProperty (easy solution,
will
>it work?)

I still don't think this is really sufficient for the sort of levels a
site wide EARL report would need, but can't see any problems, it may make
EARL servers (should they exist..) more complicated to create, but that's
all.

>* Got rid of earl:lastModified and earl:released

I still like some way of ensuring that the EARL report isn't being
applied to a fixed version, consider an Earl Annotation based server, I
add a report saying the site is broken, the author fixes it, but then
people are still making judgements based on my now obsolete report, there
needs to be some way of coping with this.

>* Got rid of confidence levels, for now (belong off of the Assertion
>anyway)

Hopefully these will be back before 1.0 is finalised, they are important
for machine evaluations.

>* Resolved to use DanBri's/TimBL's WOT stuff for DigSig

Wot is WOT?  (I did find an interesting Ethiopian Chicken recipe (too
much butter though), but I don't think that's what you mean :-)

Jim.

Received on Thursday, 1 November 2001 10:33:24 UTC