On Jun 25, 2007, at 12:15 PM, Steve Coast wrote:
On 25 Jun 2007, at 16:56, Scott Davis wrote:
At the end of the day, arguing that one is ultimately better than
the other is naive.
I didn't, I defined 'worked' as something lots of people use. How
many orders of magnitude more people use gtiles and not wms?
Wow -- it sounds like you and I are in violent agreement here. (wink)
The paragraph that you quote says essentially the same thing:
At the end of the day, arguing that one is ultimately better than
the other is naive. They represent the opposite ends of the web
mapping spectrum. GM optimizes for mass consumption (caching,
performance). To do so, they sacrifice customization. WMS optimizes
for customization. To do so, they must sacrifice caching.
Many millions (billions? trillions? gazillions?) of clicks are
handled by Google Maps every day. They have a great solution to
handle volumes of that magnitude. I wouldn't dream of putting a WMS
server in place with the expectation that it could handle volumes of
that level. Wrong tool, wrong job.
But what if you have an intranet app you want to put up? Can't use GM
-- terms of use say you can't restrict access to it. What if you want
to charge for access to your map? What if you want a different base
layer? What if you have more up to date satellite/aerial/map imagery?
What if you want a different projection? What if you want to host it
yourself? What if you want to give users the ability to turn specific
map layers on and off? What if you want to return data (WFS) instead
of pixels (WMS)?
If none of those apply to your map project, then Google Maps
continues to be the correct solution. But if even one is a show-
stopper requirement, then GM becomes the wrong tool for the job. None
of them are unreasonable requests. GM simply doesn't have a good
answer for them.
Google maps wins -- hands down -- for being an easy to use,
infinitely (?) scalable solution. It is incredibly popular. It filled
a distinct need missing from the OGC specs -- making maps bone simple
to create for non-GIS professionals. I think that all of those fill
your definition of "works". But as your needs become more
sophisticated, GM begins to fall down.
Let me borrow your technique of supplying a specific, non-standard
definition for a common word. Your word was "works". My word is
"easy". How easy is it to overlay current weather data over a Google
Map? I'd argue that it is difficult, bordering on impossible. How
easy is it if you use WMS? Trivial. (Here's a hint: <http://
mesonet.agron.iastate.edu/ogc/>)
Please keep in mind that I am mocking myself -- not you here. I'm
glossing over the huge barrier to entry that the OGC imposes in terms
of actually having to know something about mapping in order to create
a map.
You and I both are in agreement that GM is the easy (dictionary
definition), popular solution. However, once you have an
understanding of the how to construct a WMS request, which involves
understanding what a projection is, which involves specifying both my
map dimensions in pixels _and_ a bbox, then the OGC solution is
"easy" by my definition. It is no harder _for me_ to add the weather
layer than any other layer. My investment in learning the OGC spec
and all of the accompanying arcane cartographic junk just paid off in
that one "easy" transaction.
The point there was that there was this big standard for serving
maps, google ignored it, and now tiled wms is trying to catch up.
Again, you and I agree here. Google Maps is a brilliant example of
scaling maps to the masses. The OGC would be well served by emulating
their example.
You can say with 20:20 hindsight that WMS wasn't meant for what
google are doing... but I'd be surprised if during the spec process
all were agreed that they wanted to make sure large volumes were
unachievable with the spec and they'd leave it for someone else to
put large scale maps on the web.
Wow. I was neither on the board that put together the original WMS
spec nor the team of developers that created the original Google
Maps. Anything I said about the original intentions of either group
would be pure speculation on my part.
I give GM an 'A' for ease of use. I give OGC an 'A' for interop.
Nothing would make me happier than GM adding WMS support. Nothing
would make me happier than WMS being easier to use. The more lessons
each learns from the other, the better.
Google Maps scale to the masses like nothing else out there. WMS maps
are customizable like nothing else out there. Until one spec offers
unparalleled support for both, you need to know the relative
strengths and weaknesses of each and optimize towards the solution
that best fits.
-s
_______________________________________________
Geowanking mailing list
[email protected]
http://lists.burri.to/mailman/listinfo/geowanking
