[ka-Map-users] ka-Map and OpenLayers

[Tim Schaub]

Hi All-

I like the idea of a merger and look forward to more discussion.  In
general, I think pushing as much functionality as possible to the client
side is a good idea.  For ka-Map, I'm imagining that a merger would mean
all client side niceties could be retained (or improved) and all server
side functionality would be OGC spec.  So, the server side tiling,
caching, etc could be retained as long as it conformed to the yet to be
determined WMS-C specifications.  

Maybe it would make more sense to say that this is really a proposal to
split ka-Map in half - so the two part proposal would be:
1) Merge ka-Map client side functionality with OpenLayers.
2) Create a WMS-C compliant tile server out of the server side
functionality of ka-Map.

I imagine that people who used the results of #1 above (let's say it's
still called OpenLayers), would choose from a variety of tile services.
The results of #2 above (say it's still called ka-Map, or maybe ka-Tile)
would only be one of the options for a server side solution.

So, limitations come up when you want to add functionality to the client
side that relies on custom server side functionality (that wouldn't be
shared by all WMS-C options).  An example limitation would be a nice
print to PDF tool.  I think this can easily be addressed by custom
libraries or extensions.  Developing a print to PDF tool would involve
creating a tool that could plug in to the client side and call custom
code on the server side.  Though this is essentially how things are done
already, it seems a cleaner way to develop - requiring that the
application can stand alone without the custom tools.  I think others
would probably agree that ka-Map could benefit from more modular
customizations.

In terms of the other differences that Paul brings up:

> * core tiling engine is very similar, but it is not quite as 
> optimal (doesn't reuse images for instance)

I think you mean the client side tile handling here.  If so, the merged
client-side application would use the most optimal scheme.

> * overlay stuff is point/text only

Again, the merged client-side application would make use of the most
mature stuff.

> * lacking tools (layer controls, scale bar etc) and the 
> windowing stuff

I'm planning on rewriting the scale bar for prototype.js and
contributing it to OpenLayers - so that's one small piece.

> * lacking query capability

I don't know enough about WFS querying to know what the real limitations
might be.  It seems like this one deserves some more consideration.
Though, I do like the idea of having all "template" style stuff (to
handle query responses) on the client side.

> * tile caching

If a WMS-C specification is decided upon, the current tile.php
functionality could be modified to conform to that.  In developing the
specification, a better cache design might arise too.

Looking forward to more,
Tim