[Community] Random thoughts about geoprocessing

[Sean Gillies]

Yesterday, I was thinking about graphical geoprocessing model builders 
and wrote down some half-baked ideas: http://zcologia.com/news/259. In a 
nutshell, I'm curious whether we can generate data-processing Python 
code from UML by following the lead of ArchGenXML:

http://plone.org/documentation/tutorial/archgenxml-getting-started/workflow

I think an activity diagram fits the bill better than a state diagram

http://argouml.tigris.org/documentation/defaulthtml/manual/ch21.html

I received a response from Matt Perry, who has an interest in a high 
level data processing framework. I hope he won't mind if I drag his 
working proposal in here:

http://portal.nceas.ucsb.edu/Members/perry/proposal-for-an-open-source-python-gis-processing-api/

Seems to me that a UML-based modeler and such an API could complement 
each other very well. I also think if one were to drink the Zope3 
kool-aid, you could layer a processing API over PCL's feature model. 
Something like this:

     store = FeatureStore(...)
     source = store.source(typename, filter=...)
     layer = IProcessingLayer(source)
     buffer = layer.buffer(distance)
     buffer.save(file)

IProcessingLayer adapts the PCL feature source to the new protocol, 
which defines .buffer() (and others), and .save(). Python 3000 won't 
have Zope3 style adapters; IProcessingLayer would instead be a generic 
function

http://www.artima.com/weblogs/viewpost.jsp?thread=155123

Anybody aware of any other projects that are generating Python code from 
UML?

Cheers,
Sean