[Geotools-devel] Incremental improvements over the process annotations

[Andrea A. <aa...@op...>]

Hi,
a couple of days ago I grabbed Jody's process annotation prototype
and stuck it into GeoServer WPS so that I could work on it quickly
and validate it, make improvements. The result is here:
http://svn.codehaus.org/geoserver/trunk/src/community/wps/src/main/java/org/geoserver/wps/jts/

The main idea is still the same, though I've made a number of changes 
and fixes:
- respect the gt2/java coding conventions
   (it's func(a, b) not func( a, b ) )
- don't use all uppercase for local variables
- make a reusable base class that can target any collection of static 
methods
- add multiplicity in annotations
- add support for array arguments
- add support for passing a monitor in the function (if an argument of
   the static method is a ProgressListener we pass down the one provided
   to Process.execute)
- add support for pure side effect functions without return values
   (for what we know the process could make coffee and return nothing
    in electronic form)
- add version in process annotations (if missing defaults to 1.0.0, a 
describe process without a version is schema invalid)
- use converters when assigning arguments, and fail the execution if
   a conversion fails
- check the multiplicity respect before running the process

On top of that I've implemented, by copying and massaging the class
backing the geometry filter function, a factory that exposes a number
(45) of simple JTS based processes. Not interesting in GeoTools, but
interesting in GeoServer if someone wants to throw some WKT at the
server and have it transformed/analyzed.
Also intersting because a real, working implementation allowed me
to find and fix quite a bit of issues in the annotation driven
factory.

So, what do you think, good to come back in GeoTools?
Do you also want the JTS factory along with it (maybe it is of interest
for uDig?)

Cheers
Andrea