Re: Where does the EARL go?

[Just some quick notes, hope they help.]

> How make statement for whole site vs. particular
> page vs. specific element [...]

The whole site thing is tricky: RDF had "aboutEachPrefix", but it was
decided that it should be romved to a higher level in the language. CWM has
"startsWith" as a built in, but once again, it's at a higher level in the
lange. You could model a whole site in any number of ways, it's just
deciding which one is best: we can probably decide on that when we get to
it.

[...]
> It seems to me the answer is either something like RSS

RSS? We're syndicating EARL? :-)

> or something like style sheets or something like metadata
> in HTML or a file extension/mime type or perhaps all of
> the above.

Metadata linked from HTML sounds like the best solution, but we'll have to
investigate scenarios. It's not too difficult to parse HTML to strip out
link data.

[...]
> b. each page on the site has a link to the .earl file. it acts like
> an external css or script file that all pages on a site can point
> to.  then, the claim is only has to be updated once to effect
> the whole site.

Well, it's a neat scenario, but remember that an EARL file is independent
of the files which point to it. It's not as if you have an association from
a file to a piece of EARL saying "this is the conformance claim for this
file", because the EARL itself contains that information. You can say,
"here is some EARL that is related to this page, but don't take our word
for it: parse it". EARL can't exist without a TestSubject part, because it
simply wouldn't work.

> [...] there is no "type/earl" - should there be?

I reckon that we shouldn't decide at the moment, until RDF Core state
clearly the MIME type for RDF and its derivatives. I should chase that up.

For now, all EARL is going to be served as text/plain or text/xml, so just
leave the "type" attribute off. It's only advisory.

> HTML 4.01 says that new types may be created using the
> profile attribute on the head element. [4]

People tend to ignore that, sadly.

--
Kindest Regards,
Sean B. Palmer
@prefix : <http://webns.net/roughterms/> .
:Sean :hasHomepage <http://purl.org/net/sbp/> .

Received on Friday, 19 October 2001 13:58:25 UTC