The main advantages of LSIDs are pretty much what you have listed. The main ones being the separation of the Identifier itself and the resolution protocol
and the encouragement to avoid “plain” urls that are so easily broken. This was obviously much more of an issue 8-10 years ago when we started discussing the use of GUIDs in biodiversity informatics. These days however, with the move towards restful architecture
and web resource based mechanisms (eg avoiding urls like
http://example.org/birds/moa.html, which doesn’t make a very nice ID), I feel the approach for creating web resources and urls has improved and result in less broken urls anyway.
I think I agree with Rod here in that it is probably best to adopt the benefits of LSIDs but use HTTP URL/URIs.
Kevin
From: tdwg-content-bounces@lists.tdwg.org [mailto:tdwg-content-bounces@lists.tdwg.org]
On Behalf Of Roderic Page
Sent: Tuesday, 19 June 2012 3:39 a.m.
To: Julien Cigar
Cc: TDWG Content Mailing List
Subject: Re: [tdwg-content] LSID Resolution Service using DDDS/DNS
Hi Julien,
On 18 Jun 2012, at 15:51, Julien Cigar wrote:
Hi Roderic,
Thanks for your reply.
But I've almost the same feeling. The more I read about LSIDs, the more I feel that there are no real huge benefits over URLs, .. although it looks like different concept to me: I see more an URL as the location of the resource, and the LSID the identifier
of that resource.
DOI can do the same thing, or we can design URLs to act much the same way (e.g., by using consistent "slugs" across domains, such as Wikipedia topics or ISBNs).
It's frustrating because at first glances all the models and the "to be compliant you must ..." rules that come with the spec seemed pretty
Some advantages I see compared to URLs:
- the separation of data and metadata (through the .getData / .getMetadata methods and the WSDL files), .. although you could achieve more or less the same by playing on the Accept header of the HTTP request
The metadata/data distinction is often messy. What is metadata and what is data? Is the metadata about the data, or about the LSID itself, etc.? Cue much metaphysical discussion (sigh)
- the versioning
To a first approximation nobody uses versions. I have never seen anyone use a version of a GenBank sequences, for example. In general people people want the latest version. One can also create versioned identifiers using other schemes if
you really must.
- the resolution services/discovery services.. although there are a lot of "assuming that..." in the spec
They didn't happen...
- protocol independent
I think HTTP pretty much won that argument...
- other things ... ?
LSIDs do have some nice features, but if we step back we see that it's got zero traction outside our community. It's either URLs or DOIs (the later are gaining traction in the data citation world, and are obviously already established in
academic publishing). There's a bunch of reasons LSIDs failed to take off, partly technical (they are too hard for most people to get working properly), partly social (pretty much no LSID serves content we can actually do something useful with).
I'm being a little flippant, but personally I think the last thing we'd want to do as a community is continue to flog a technology that has zero uptake outside biodiversity informatics. it was a nice idea, we tried it, didn't work out too
well, let's move on...
Regards
Rod
Best regards,
Julien
On 06/18/2012 15:15, Roderic Page wrote:
Hi Julien,