Re: Straw Poll (ISSUE-1)

I agree with Gregg that 1 will not work. In particular, I believe it will
fail for terms created using schema.org's extension mechanism [1].

I believe that a registry is the best way to accomodate the flexibility of
microdata vocabulary specifications while also providing consistent URI
generation. I believe that this can be effectively crowd-sourced so that
vocabulary publishers don't have to go to extra effort.

While vocabulary based parsing seems an untenable solution, Richard Cyganiak
has a proposal for how 2 URI generation patterns could be defined for
vocabularies to follow; vocabulary based or itemtype based. This would allow
processors to just be coded to handle either of those two. Then the registry
could just designate which pattern a vocabulary uses. The key point
here: *Processors
would not need to be updated for vocabularies*.

Richard has done a lot of good thinking about this, so I asked him to write
it up [2]. I know it's still very much a work in progress, but I think
developing this idea further could accomodate the flexibility of microdata
without introducing inconsistency.

-Lin

[1] http://schema.org/docs/extension.html
[2] http://richard.cyganiak.de/2011/10/microdata.html#vocabs


On Tue, Oct 25, 2011 at 1:59 PM, Toby Inkster <tai@g5n.co.uk> wrote:

> Sorry this response is post-Tuesday-AM, but my preference would be:
>
>        1. all processors use the same (default) mapping for all
>        vocabularies
>
> Why? #3 just sounds bad, and I think as time marches on, #2 would
> eventually become #3, as some processors would be updated to cover
> newer vocabularies while others stagnated.
>
> --
> Toby A Inkster
> <mailto:mail@tobyinkster.co.uk>
> <http://tobyinkster.co.uk>
>
>
>


-- 
Lin Clark
DERI, NUI Galway <http://www.deri.ie/>

lin-clark.com
twitter.com/linclark