[mapserver-dev] RFC: GMaps API for mapserv

[Paul Ramsey]

>  Yes, I agree with the fact the metatiling is out of scope. I think the
> point that I did not make so clear is that assuming someone wants to do
> metatiling, say using some script or tilecache or whatever. Is there
> anything in the implementation or design that would prevent one from
> generating metatiles and chopping them up via a script and storing them
> somewhere outside of mapserver? If yes, is removing that within the scope of
> things or not?

The main thing preventing using this as a metatile server is the
assumption of 256x256 tiles.  Basically (and this relates to Paul
Spencer's request to name it mode=tile) saying "gmap" implicitly sets
a number of conditions:

- The CRS will be "spherical mercator"
- The bounds be global in extent. The top level will have 1 tile
(http://mt.google.com/mt?x=0&y=0&zoom=17)
- The "levels" will run from 17 downwards
- The y axis of the tile coordinates will run from top to bottom
- The tiles will be 256x256 in size
- Each zoom level will be related to parent and children by powers of two

(Digressionary question to anyone who knows or can guess: why would
google choose 17 for their top level and count down, rather than 0 and
count up?)

A generic "tile" mode would presumably also serve other schemes, of
which there could be many. The most obvious is lon/lat with a similar
set of image size and pyramiding constraints.  The worldwind page
doesn't describe their top level tile or their numbering scheme,
unfortunately, but I'm sure we could figure that out. We could go with
their scheme, and something like this:

http://server/mapserv?map=/foo/bar.map&mode=tile&interface=gmap&version=1&x=0&y=0&level=17

And then also serve the two worldwind variants:

http://server/mapserv?map=/foo/bar.map&mode=tile&interface=map&version=1&x=0&y=0&level=0
http://server/mapserv?map=/foo/bar.map&mode=tile&interface=worldwind&x=0&version=1&y=0&level=0

And whatever the CRS/tile layout variants are for the other commercial engines.

On forcing "PARTIALS TRUE" would it be fair to do that only if
PARTIALS was unspecified?

On rendering large and clipping, can I add that to the RFC as "phase
2" :)  Gotta walk before I run. But no philosophical objection on my
part.

P