[OSGeo-Standards] [REST.SC] Was: "file" formats. Is: GeoWeb

[Rushforth, Peter]

Hi Pat,

There's lots to agree on.

> 
> > 
> >> Commercial search
> >> engines are indeed the first place people typically use to find 
> >> information, so the technology should be better leveraged to allow 
> >> users to search for geospatial information with appropriate 
> >> semantics.
> > 
> > +1
> 
> Indeed
> 
> > 
> > Search engines are enabled by the fact that they can crawl 
> documents.
> > Having a declarative approach (hypermedia) to spatial data, 
> instead of 
> > an interface approach, will allow the evolution of standard media 
> > types which can be crawled and searched.  ISO 19115 is a 
> possibility, 
> > but there are others which are more data centric thatn metadata 
> > centric. e.g. Atom
> > 
> >> 
> >> On the topic
> >> of a pragmatic interface, I believe we should explore the 
> development 
> >> of a human-accessible HTML layer
> > 
> > That's what I'm advocating, except it HTML should be the client of 
> > data services.  What it points to must be declarative, to avoid the 
> > "R" word for you.  What is consumed by the HTML layer should be 
> > embodied in a standards-track internet media type.
> 
> This is the exact approach that Facebook has taken (which we 
> are now following) When you publish objects or products, you 
> do so via an HTML page that contains specific HTML meta-tags 
> (RDFa). 

I think this is workable.  But the media type is always text/html.  
How does a client request html containing the specific meta-tags that 
represent a concept?  Or does html just keep on growing with no 
definition and no need to modify the standard?

In any case, I'm advocating for adding a tag for map to html in a 
slightly more formal manner than tunnelling RDF inside of text/html.  
The geo community needs this, and it is long overdue.

> Please see Facebook OpenGraph.

I'm not on facebook, don't plan to be either.  The Web, on the other hand, I am on.
But sure, I'll have a look at that spec.

> These tags can then be crawled by any bot include google-bots 
> for indexing.
>  No manual cataloguing required.

I think that is ideal, except for the media type stasis.

> Does this scale? YES.  That's a fact.
No argument there.

> Are we arguing about the nature of REST? NO. 

I sure don't want to get into that debate again.  But I want standards
based on the reality of Web architecture, which is not 
esoteric or far-fetched.  It actually works, when you look at it. 
*That* may be the magic ;-).

> There is nothing magical or threatening about level 5

We need declarative, standards-track internet media types supporting geo.

> Do we need prototyping?  YES of course.

+1  "Rough consensus and running code" is the mantra of the IETF.

> Do we need to abandon existing OGC Services and efforts? NO, of course.

+1  The community is our strength, actually.  

Not only should standards be agreements on what works, but they should also serve to 
educate the reader on technology 'best practice', if you like to call it that.

That said, we as a community need to be a bit open to influence from 
the success factors of other communities, notably IETF and the W3C.  
The risk of not being so open is that we may end up creating a little 
geo-ghetto of our own making, hidden behind our "standard interfaces".

This is why I am advocating for integration of OGC concepts (WMTS, GML, Simple features, etc)
into IETF standards (Atom, AtomPub), as well as W3C standards (hypermedia,html,  etc).

Regards,
Peter