[tdwg-guid] Handle System considered not interoperable with standard WWW and SW applications

[Ricardo Pereira]

    Markus,

    We really value your opinions, as well as those of everyone else. We 
definitely don't want to keep you silent or force you to use a 
technology that you think is inappropriate. However, there is a clear 
divide between LSID and HTTP URI identification schemes (maybe even 
Handles?!?), and we need to get past that issue to progress.

    If we don't reach consensus about which single identifying scheme we 
should all use, I see no other alternative than letting people use the 
identifying schemes of their choice and let the systems fight it out 
(come again??!?! what did I say?!?!)

    Well, this idea may seem radical at first, but really the 
implications are not that relevant as long as-

i) we only consider LSID using HTTP proxies and a (yet to be specified) 
identification scheme based on HTTP URIs.
ii) we representat objects in the same way regardless of identification 
scheme in use.

    In either case (LSID with proxy and HTTP URI), the clients will 
always come across HTTP URIs when navigating through linked data, and 
thus will always use HTTP to deference the URIs and get to the object 
metadata. The metadata in turn will be expressed in the same way 
regardless of which id scheme you use (as per rule ii above). Thus, the 
identifying scheme used  to identify an object will be almost completely 
transparent to clients.

    There will be a few details that will make things a little different 
in one case or the other (only LSIDs will let you get data for an 
object, and so on). Those will the main reasons why one would choose one 
scheme over the other. In 200 years time, we will know who were right 
and who were wrong. But until then, we all go our separate ways, but 
keeping our systems interoperable.

    We are now drafting an LSID Applicability Statement in which we 
specify the rules that must be followed by anyone using LSIDs in our 
community. That document, however, won't say that anyone in our 
community must use LSIDs. So if you want to propose an alternative 
identifying scheme, you suggest you do the same: draft a detailed 
specification for your new scheme, submit it to the TDWG standards track 
and put it up for review by the community. Otherwise, if we don't have a 
detailed spec for HTTP URIs to refer to, we will always get stuck on 
those what-if questions. And please follow the 2 rules I outlined above 
so that we keep both systems interoperable.

    The same applies to handles. If anyone wants to use handles, he or 
she must draft a detailed applicability statement defining rules for 
using handles in our domain and submit it for review. But again, I would 
advise against handles because they won't be as interoperable with the 
other two schemes for the reasons I outlined in an earlier message 
(incompatible content-types).

    I hope this helps.

    Best regards,

Ricardo



Markus Döring wrote:
> Donald, Ricardo,
>
> I fully agree with you and Rod that we need to go forward as fast as 
> possible and dont need yet another discussion. If everyone else is 
> pleased with LSIDs I will keep silent, promised. LSIDs are better than 
> nothing. But if we would use URLs we could go a lot faster 'cause its 
> so much easier. Especially now after all the lessons learned.
>
> Still some comments below inline
> Markus
>