[geotk] News on current work

[Martin Desruisseaux]

Hello all

This has been a very long while since I didn't gave any new about 
current work, so I though that an update is due.

The 3.20 release has been delayed for a very long while (almost one 
year!!), mostly because of extensive work on the support of NetCDF 
format. While we use the UCAR library for reading the binary file, there 
is a significant amount of work to do for connecting that to the Geotk 
metadata, referencing and coverage frameworks. They were also lot of 
work done on testing with more authoritative tests (more on it below), 
and new features in the referencing engine which are (I think) unique to 
Geotk.

I'm trying to brings the NetCDF support to a stage where it is complete 
enough before to release Geotk 3.20. This willing to avoid release with 
half-backed solution is the reason for the delay.

More specifically, work done in the last months include:


Metadata
--------
One noticeable improvement is the mapping of NetCDF-CF convention to ISO 
19115 metadata: 
http://www.geotoolkit.org/apidocs/org/geotoolkit/metadata/netcdf/NetcdfMetadata.html 
(not exhaustive).


Referencing
----------
We added the projection parameter names for Proj.4 and NetCDF. By doing 
so, it became more and more obvious that the various parameter names 
where in state of chaos and needed to be organized. The result is 
documented here: 
http://www.geotoolkit.org/modules/referencing/operation-parameters.html 
(note that this page was generated by GeoAPI, so similar pages can be 
created for other projects).

Geotk is able to compute transform derivative for almost every 
projections. As far as I known, this is a feature unique to Geotk - 
thanks to a good mathematician in our team. Derivatives are useful for 
calculation of Bezier curves, computing Grid Coverage resolutions, fast 
raster reprojections, faster and more accurate Envelope reprojection, 
and more.


Testing
------
We created one year ago a new GeoAPI module, called 
"geoapi-conformance". We are slowly relying more and more on that module 
for the tests. Those tests are created with the same focus on quality 
than the Geotk library. Furthermore we run the tests on other libraries, 
especially Proj.4. Analysing the tests successes of failures with 
different libraries make us confident about the tests robustness. Those 
tests could hopefully be leveraged by any project that which to do so.

Generic tests: http://www.geoapi.org/geoapi-conformance/apidocs/index.html
NetCDF-specific tests: 
http://www.geoapi.org/geoapi-netcdf/testapidocs/index.html


     Regards,

         Martin