[Gdal-dev] Re: gdal 1.4.1 python 2.4 windows executables

[Emilio Mayorga]

> > I am interested in making "core binaries" for win32 part of the funded
> > GDAL maintainer workload, but it isn't at all clear that this will include
> > Python packages.
>
> I sure hope it will.
>
> The easiest thing for Windows users is to have binaries for each
> release, for each recent version of Python (I think the two most recent
> is good enough -- i.e. 2.4 and 2.5). However, barring that, at least is
> should be all set up to build with distutils. That way at least anyone
> with the right compiler can "just build it". Indeed, then they could
> donate the binaries back to the project.

I'd like to add myself to the (small?) chorus. It'd be terrific to
have GDAL Windows binaries for recent versions of Python (I agree that
2.4 & 2.5 is good enough). FWTools is great, but stand-alone GDAL
packages that can be integrated with the available Python installation
would provide the most flexibility.

To illustrate... I'm currently developing a package of models of
global river nutrient export to the oceans
(http://www.marine.rutgers.edu/globalnews/); I'm taking the work
previously published by our work group, including implementations in a
set of Excel spreadsheets, and integrating them into one package that
uses Numpy to provide Matlab-like, clean access to the model equations
and data input/output. GIS pre-processing (raster to tables as "zonal"
statistics for watersheds) is done behind the scenes. Our partners
(scientists, new to Python) need to be able to install this package on
their systems, with minimal hassle; on Windows, currently that's just
Python and Numpy, very simple. For the GIS part I chose to use ArcGIS
geoprocessing for expediency. StarSpan
(http://starspan.casil.ucdavis.edu/) unfortunately wasn't an option
b/c it's missing a bit of functionality, is apparently not
straightforward to install on Windows, and seems orphaned at this
time. FWTools on top of a separate Python install would've created too
much complexity and the Numpy bindings are not there yet (?).

Python with Numpy, matplotlib, SciPy, and GDAL (with the "Next Gen"
Numpy bindings), all easy to install: wow, that'd be one sweet,
compelling environment for earth science computing/"geoinformatics"!

Unfortunately, I'm not offering to help create these executables
either. I wish I could send money your way, but I seriously doubt our
funders appreciate the benefits of this sort of support for external
developers that's not very specific to the goals of our project.

Thanks for all the great work. Cheers,

-Emilio Mayorga
Rutgers, The State University of New Jersey
Institute of Marine and Coastal Sciences
71 Dudley Rd.
New Brunswick, NJ, USA 08901-8521