Re: Banging the identity drum again, this time REST-style

On 26 Nov 2010, at 5:01 AM, Murray Altheim wrote:

> On 26/11/10 04:47, Andrew S. Townley wrote:
> [...]
>> I hadn't realized this before, but I just saw a recent post to
>> xml-dev[1] which, in the words of Rick Jelliffe, states:
>> 
>> 	"ISO officially does not allow reference implementations"
>> 
>> Perhaps this is an issue, or perhaps not.
> 
> No, I think not. "Allow" is a funny choice. Do they *prohibit*? or just
> not endorse?  ISO 8879 has as its reference implementation James
> Clark's sp toolkit. That's partly just because of James' reputation
> I suppose but anybody that could demonstrate an application that
> acted as a reference implementation could probably sell a community
> to treat it as such. I don't think there's any ISO endorsement really
> necessary -- it's an act of acceptance by a community, and that probably
> varies by community.
> 

Letting the horses speak for themselves:

[Rick Jelliffe] http://www.stylusstudio.com/xmldev/201011/post60150.html
[R. Jelliffe] http://www.stylusstudio.com/xmldev/201011/post70150.html
[Michael Kay] http://www.stylusstudio.com/xmldev/201011/post80150.html

I can see both sides here, but I can certainly emphasize with Michael's point about maintaining bug-compatibility with a reference implementation.  The unintended behaviors of a reference implementation might get in the way of people actually optimizing things for a particular purpose.

I guess I'm more in favor of a really well validated specification (or set thereof) with some equally well validated, yet non-normative accompanying documents like Primers, a la XML Schema that give additional insights to both users and implementers of the intents and rationale behind a specification crucial for understanding it correctly, but which don't belong in a formal specification aimed primarily at members of a particular community who most likely don't need a rehash of the background and rationale.

Having written a couple of these, I do know from first-hand experience that if you want wider adoption and understanding, there's a happy medium between the two approaches.  However, it's a fine line at times.

[...]
>> I have some ideas to explore on my own in this space, but they
>> aren't getting a lot of personal CPU cycles based on what I'm
>> trying to do by the end of the year either.  I don't think I'm
>> alone with this, somehow... ;)
> 
> I've still got one project I'm hoping to explore that could go
> public.

Looking forward to seeing that! :)

[...]
>> As Yoda says, "You must unlearn what you have learned."
>> Letting go of some pretty fundamental software development
>> "truths" seems to me to be necessary to really "get" Topic Maps.
>> Unless we can show why this painful process is actually worth
>> the effort, only a few brave (or, in my case, foolish) souls
>> will dare start the journey.
> 
> Well, it is a paradigm shift just to graph theory for a lot of
> people. When one adds in the variety of new concepts in Topic
> Maps it's perhaps not surprising that so few people grok even
> the basics. A lot simply think it's a different syntactical
> representation of RDF, just another graph markup.


Like it or not, most things that have achieved mainstream awareness have passed the "O'Reilly test", e.g. there's a book from them on a particular subject.  The precursor to that is normally a widely available open source implementation that people can start playing with and actually use in their projects without a licensing cost.

While people's opinions of it vary, Ontopia now meets the second criteria.  Now all we need is someone to use it in a way to solve an enterprise problem as a complement or replacement to a "standard" approach, them to publish a lot about what they've done and get it picked up by an influencer like an open-minded analyst or a developer/architect with a large following (I can think of several, but Fowler's probably at the top of the list).  Then, that buzz will allow either O'Reilly or Apress to think it will actually sell some books and get someone in the community approached to write a manuscript.

Time and effort involved: ~1-3 years, depending on how fast people write... ;)

I want this to happen for the simple reason that I'm really very lazy.  I introduced Topic Maps and Topincs to an organization providing a large amount of public information via PHP+MySQL (LAMP) so that it would be easier me to integrate their information on an ongoing basis.  We'll see how it goes, but they seemed interested. :)

Cheers,

ast
--
Andrew S. Townley <[email protected]>
http://atownley.org