[Taxacom] AMNH, spiders, and LSIDs

[Ricardo Scachetti Pereira]

    Hi all,

    In theory, there is nothing wrong with issuing LSIDs without 
resolving them. The only hard and fast rule is that you cannot assign 
the same LSID to more than one object.

    There are a couple of practical problems, however:

1) People will expect to be able to resolve the LSIDs they come across. 
If you (or an automated agent) get a reference (LSID) to an object, you 
(or the agent) may want to know what it points to, and you'll do that by 
trying to resolve it. Applications will try to resolve them directly 
using the LSID resolution protocol, while humans may try to resolve them 
using a proxy, like http://lsid.tdwg.org/. If your LSID doesn't resolve, 
that's a usability problem, much like the dreaded 404 error - page not 
found. We've experienced an example of that when Rod and Kevin tried to 
resolve the LSIDs given in the beginning of this thread.

2) There are also technical implications associated with the choice of 
each part of the LSIDs one mints that affects the deployment of 
resolvers. The best way to take these implications into consideration is 
to try to deploy a resolver *while* you are defining the parts of your 
LSIDs. If you define your LSIDs long before you deploy your resolver, 
there will be technical consequences that you will have to live with 
(forever!?) when you deploy your resolver. In a few cases, you may not 
even be able to deploy the resolver at all (if you make a bad choice of 
an LSID authority identification - the domain name in the LSID, for 
example).

    That's why we recommend in the TDWG LSID Applicability Statement 
that LSIDs should be resolvable (recommendation #6). You can find that 
document at http://www.tdwg.org/standards/150 (click the "Download" link 
- it's a MS-Word document for now as it is under review in the TDWG 
Standards Track.)

    In the next few days we are about to release a set of LSID authority 
implementation guides that addresses some of the issues I mention above. 
Those documents will help people mint LSIDs but they won't replace good 
hands-on experience that you get by deploying your resolvers.

    I hope this helps.

    Cheers,

Ricardo


Roderic Page wrote:
> I take your point. But, if I was to publish a web site with lots of  
> URL links, none of which actually worked, would anybody take me  
> seriously? I suspect not. I wonder what the wider community is to  
> make of LSIDs which don't actually do anything ... might they wonder  
> what the fuss about LSIDs is all about?