[Mapbuilder-devel] Re: [ka-Map-users] Hello from the Mapbuild
Mike.Adair at CCRS.NRCan.gc.ca
Thu May 19 09:03:14 EDT 2005
You should definitely invite me over to give another demo, things have
evolved quite a bit since then.
If you guys are serious about doing an AJAX client, you should look at
Mapbuilder before you spend a lot of time and effort developing another
library. We have been implementing AJAX since long before someone came up
with a cute name for it and I think we would both be much farther ahead.
Some more comments inline.
> Design goals ... um ... what are those? :> This was
> AJAX and grew into something we felt pretty decent about from
> a performance/usability point of view.
> The original design goal, then, was to produce something that
> took maximum advantage of the natural advantage that the DOM
> and properly structured HTML/CSS can provide for performance.
This is something we've always realized and it's the reason why Mapbuilder
works so well IMHO.
> The source released today is pretty much that initial effort
> and I think it reflects the lack of any real forethought.
> The whole thing is more or less a toy application since it
> doesn't do anything useful.
> The newer version that will be available in cvs in the next
> object library with the intention to expose an API to
> technology. I am attempting to abstract the core rendering
> engine (tiling) from the tools that do stuff like navigation
> by using a custom event mechanism that links the tools
> together. And I hope to incorporate other mechanisms like
> querying and location-based services too.
> Also, I would like to position ka-Map to be embeddable in
> other technologies like Chameleon, GeoServer etc if that makes sense.
This is the same audience we are targeting although more embedding into
plain HTML and CM systems. (GeoServer is distributed with a Mapbuilder demo
> I'm not entirely sure how well this technology would work for
> MapBuilder since it relies on some sort of tile caching for
> performance. OGC protocols are not particularly known for
> performance, and adding an additional requirement for
> requesting separate tiles would be a killer in my mind. I
> think it is more likely that ka-Map will benefit from the
> things you guys have been doing.
Our library supports the OGC standards, but it is not limited to that. One
of the key features of Mapbuilder is that it is very extensible and
configurable, and very modular so all functionality is defined by objects.
For the ka-map tiling behaviour, it would require a new Model object and a
method on the drag pan handler.
> I hope that other developers (like you :)) will take a look
> and perhaps find something useful or contribute changes back.
More information about the ka-Map-users