[tdwg-tapir] GUID support in TAPIR (or BioCASe)

[Renato De Giovanni]

Hi Ricardo,

I'll try to answer your questions...

> 1) Does the protocol or the implementation explicitly handle any form 
> Globally Unique Identifiers? Does it adopt a specific technology (DOI, 
> Handle, POI, LSID), or does it provide support for any GUID technology. 
> If so, what are the requirements for using a new technology?

No, TAPIR doesn't handle those technologies in terms of having any 
sort of implicit resolution mechanisms or any other specific 
behaviour. TAPIR can "serve" GUIDs if they are part of a conceptual 
schema and if providers store the corresponding values in their local 
databases. However, they will be treated just as simple strings, both 
in filters and output models.

> 2) Are there any required fields/concepts on the application profiles 
> (ABCD for example) that are specifically designed to accommodate GUIDs?

As far as I know ABCD has "UnitGUID", "DatasetGUID" and URLs for 
MultimediaObjects. The new version of DarwinCore has a 
GlobalUniqueIdentifier concept.

> 3) Does the protocol provide any specific resolution mechanism (i.e., 
> returning a single object based on an ID passed on the request)?

TAPIR could work very nicely with PURLs just by making use of  
standard Web technologies. For instance, ABCD and DarwinCore 
providers could define a custom search template that would require 
the catalog number as a parameter and would return an RDF output 
model for the respective object. In this case, a request could look 
like:

http://example.net/provider.cgi?op=view&template=someuri&cnum=123

Which could be easily wrapped as a PURL:

http://purl.org/provider/123

So the representation of that object could be retrieved from any 
browser or from any script using standard HTTP libraries.

Providers would obviously be free to decide about the output model 
for their objects (an RDF representation, or ABCD, or plain 
DarwinCore...), and if the template should take an additional 
parameter for versioning (or maybe the response could just include a 
changelog), and if it would be desirable to have two different search 
templates (for data and metadata) and so on.

Resolution of any other kind of GUIDs would require additional 
software considering only a pure TAPIR implementation. Unless the 
implementation decides to offer additional services as part of the 
package.

> 4) Any other aspects of the protocol or implementation regarding GUIDs.

Only minor things like recommending GUIDs to identify 
"relatedEntities" in TAPIR metadata responses. I guess we will 
probably recommend PURLs for templates, output models, concepts and 
conceptual schemas. 

Hope this helps...

Best Regards,
--
Renato