Re: LDP Interoperability

Alexandre Bertails wrote:
> On 11/22/2012 01:19 PM, Nathan wrote:
>> Alexandre, All,
>>
>> I've seen you mention LDP interoperability many times, I agree it's very
>> important.
>>
>> A) Every LDPR is compatible WebID 1.0.
>> B) Every WebID 1.0 Resource is compatible with LDP.
>>
>> You want (A), but you are asserting (B) in order to argue for a
>> #fragment constraint.
> 
> Not exactly.
> 
> 1. The LDPR would concern only the WebID Profile. With hash URIs for
> WebIDs, the distinction (aka. avoiding http-range-14 for WebID) would
> be made by design.

Could be made by design. This is something I understand in minute 
detail, and dearly want, I strongly and vocally advocate using #frag 
URIs and getting rid of 303 URIs wherever and whenever possible.

The hr14 issue is not a technical one, what you want to do is avoid 303 
URIs for WebID - but what you are really doing is bringing the 
httpRange-14 debate in to WebID - it's a social debate which has no end.

The fact is, people already have 303 http URIs which denote Agents, and 
which dereference to Turtle (and other formats). If these are not 
WebIDs, what are they?

By discounting 303 URIs, only two things can happen:

1) This debate goes on and on eating up most of our time whilst the 
majority of implementations be non conforming by choice.

2) Another (competing) 303 friendly WebID-like Specification is defined 
and used by those who will not back down and use #frag URIs.

The only way to avoid both of those horrible paths, is to not exclude 
303 URIs.

IMHO, the best thing we can do is write the entire WebID 1.0 
specification for #frag URIs, and simply remove the "MUST/SHOULD use a 
#frag URI" - this way we aren't excluding them, but we are firmly 
sitting ourselves on the #frag side, if people want to use 303 URIs then 
they're on their own and can work out the nuances and drawbacks entailed 
themselves. But we can't exclude them. Hopefully LDP will take the same 
approach.

Best,

Nathan

> 2. The relationship with LDP is basically: don't paint yourself in the
> corner by making a choice that would introduce some incompatibility
> with LDP. That's because 303s and HTTP POST/PUT/DELETE don't play nice
> together.
> 
> Alexandre.
> 
>>
>> WebID interoperability entails that we ensure (A) is true.
>>
>> (B) is an LDP interoperability issue, and should be addressed in the LDP
>> WG.
>>
>> Best,
>>
>> Nathan
>>
> 
> 

Received on Friday, 23 November 2012 10:23:37 UTC