Re: includehosts mandatory?

-1 from me too.

The purpose of making includehosts mandatory was to grant the 
specification of "safe" IRI set definitions (i.e., it is unlikely that a 
content provider wishes to state that "all resources with an IRI path 
starting with /foo are red", although it is a valid statement). Since we 
have introduced the idea of using XSLT for validating POWDER documents, 
and, possibly, as a support to DR authors, IMO includehosts can be optional.

Cheers

Andrea


Smith, Kevin, (R&D) VF-Group wrote:
> 
>  >> You're saying that we already need a 2-step validation and that we
> can therefore check that an IRI set definition is not empty without
> having to mandate <includehosts>? That would be good I think.
> 
> Exactly, so -1 from me for a mandatory includehosts.
> 
> -----Original Message-----
> From: public-powderwg-request@w3.org
> [mailto:public-powderwg-request@w3.org] On Behalf Of Phil Archer
> Sent: 05 June 2008 12:21
> To: Public POWDER
> Subject: Re: includehosts mandatory?
> 
> 
> 
> 
> Smith, Kevin, (R&D) VF-Group wrote:
>> Hi Phil,
>>
>> My understanding was that if we support other identifiers (ISAN, ISBN)
> 
>> etc. then <includehost>s would not be relevant for those, only for 
>> those retrievable over the Web.
> 
> True - but POWDER-BASE only has <in/excluderegex> with no other child
> elements allowed for <iriset>. Things like ISAN numbers would exist in a
> a format that is not POWDER but that can be transformed into
> POWDER-BASE. Therefore the question about <includehosts> only applies
> specifically to POWDER and not BASE or S.
> 
>> NB a thought to park for now: so you could describe the 
>> characteristics of a bunch of SMTP addresses using POWDER, couldn't 
>> you? That seems quite powerful.
> 
> Yes indeed, and POWDER-BASE can do that, therefore conformant POWDER
> Processors can do that which is, I think, at the heart of the Stasinos
> plan.
> 
>>>> It gives us a way to ensure syntactically that an IRI set is never
>> empty
>> It seems like we will need two-step validation due to some features 
>> absent in XML Schema, so this could be picked up by an additional 
>> check (we've mentioned using XSLT as a way to enforce business rules, 
>> and also to create a neat HTML page teling you what the DR says in 
>> plain English).
> 
> You're saying that we already need a 2-step validation and that we can
> therefore check that an IRI set definition is not empty without having
> to mandate <includehosts>? That would be good I think.
> 
> P

Received on Thursday, 5 June 2008 13:02:12 UTC