[openbiblio-dev] bibjson -request for feedback

Jim Pitman pitman at stat.Berkeley.EDU
Tue Feb 21 19:20:32 UTC 2012


Peter Murray-Rust <pm286 at cam.ac.uk> wrote:

> > Without reading the details I am in favour. We develop our chemical
> dictionaries using namespaces, the crystallographers are starting to do the
> same, etc. Namespaces are 15 years old - people are starting to feel
> comfortable with them. (This is not a joke - it takes that sort of time).
> The JSON-LD spec is still in draft. Presumably:
> * there won't be lots of software yet?
> * it could get pulled. In which case we have to support the software
> ourself. That's probably just about possible

I agree, it just doesnt look that hard.

> * does it degrade gently? what do we lose if there is no JSON-LD processor?

I'm guessing we can map bijectively to JSON-LD from use of the Linkage
Schema in the old spec, provided our records do not get more deeply nested than
we are supporting already.  It may be more pleasant for many users  to work with the
naive format, and let parsers do the heavy lifting to get to JSON-LD.

> I think the risks are much smaller than the benefits.

Mark, can you comment on the risks? Main one I see is 
1) short term loss of dev time mapping the present spec to LD (short term, can you estimate that), 
2) long term friction from unpacking the LD to get at the data programmatically for various 
purposes.  
If mappers from JSON-LD to the python objects we are already using are fast,
 then 2) maybe not a problem. But I think we should demo that. Any other risks?
I do see the benefit of hooking up with another community developing tools we may be
able to leverage.

--Jim




More information about the openbiblio-dev mailing list