Our Friend, the Atom
I too think so. Dear geospatial community, please do not propose yet another special protocol until you can demonstrate that Atom won't work.
I too think so. Dear geospatial community, please do not propose yet another special protocol until you can demonstrate that Atom won't work.
Comments
Re: Our Friend, the Atom
Author: Allan
OK. I'm excited. But I can't shake the feeling of square peg and round hole. APP is brilliant at what it does, but like any decent 21st century spec, it only does a very little bit. APP is a great pattern for dealing with feed-like data. I.e. "here's a feed of the ancient coin discoveries" or "here's a feed of real-time earthquakes". But it's not a great pattern for dealing with "get me the paved roads in california." For that we're going to need another pattern.Re: Our Friend, the Atom
Author: Allan
One quirk baked into APP is that you MUST return things ordered by time (APP 10.1) if you're returning partial collection lists. Another quirk is the notion of categories. These are the *only* metadata elements treated specially in APP. Again, I think APP is good for what it is. Maybe the best thing for the GIS community is to take a long, hard look at Section 5 of the APP draft and come to grips with what REST is. And, to be slightly pedantic - Atom is a format. Atom Publishing Protocol is a protocol.Re: Our Friend, the Atom
Author: Sean
Allan, why isn't APP -- or something from the same abstract super group of protocols -- a good pattern for getting representations of entities in a spatial context? I don't accept "spatial is different because it is" anymore.Re: Our Friend, the Atom
Author: Allan
Note that I didn't say APP is not "a good pattern for getting representations of entities in a spatial context". I said it was not a great pattern for a certain class of queries. Mostly what I'm trying to say is that as a pattern, it might have its limits. I'm not against trying to bump into the limits. It's a gut feel. I'm also not saying that spatial is different. I would have the same reservations about using APP to find all the sentences in all the books at Amazon that have the words "if" in them where the preceding sentence had the word "the" in it. To do that you'd (a) have to define a query protocol that APP does not give you, and you can imagine that the result set could be quite huge. Most importantly, the things that APP makes "first class" - time and categories, are irrelevant to the problem of searching a text corpus. Again, I'm not saying APP is not a good protocol. I'm saying I get the sense that it should properly be viewed as a component, not as the salvific balm to all of geospatial's prior sins.Re: Our Friend, the Atom
Author: Allan
I'm actually quite enthusiastic about the winds of change that are blowing through the geo hacker community. The current debate/investigation of REST, things like tilecache, featureserver, geojson, rtree, etc. will, in combination, blow away layers of crud.Re: Our Friend, the Atom
Author: Sean
Well, there's never a silver bullet, but switching to protocols that are of the web is a big step in the right direction.