[gdal-dev] New Features, Future Steps

[Frank Warmerdam]

Folks, 

I have checked in substantial changes to GDAL, amoung other things:

o Incorporating support for metadata (read), via the GetMetata() API on
the GDALDataset and GDALRasterBand. For now only TIFF returns a few
things, but I am expecting to return rich calibration metadata from
some CEOS datasets.

o Incorporating the ability to read GCPs from a dataset. For now there is
no code in GDAL to build a polynomial model from this, but I may add this
at some point. Currently only returned from TIFF, but support will be 
added for CEOS and some Atlantis formats. 

o I have modified the shared library to be call libgdal.1.1.so so that 
-lgdal.1.1 can be used to link against it without hardcoded paths. 

I have done a test build on Windows, and after a few fixes it works there too.

Future issues:

o I am thinking about switching to use libtool to build the shared library
and adding an install target to install libgdal into a fairly standard
way. I recently incorporated use of automake into PROJ.4, and it worked
well, but because of the multiple directory nature of GDAL I don't think
it is well suited to using automake. 

o There is the issue of whether to separate the "OGR" vector support, and
the GDAL raster support into separate libraries. I have been knashing
my teeth about this for some time. The problem with separating is that
substantial components are currently shared, notablely CPL (gdal/port), 
ISO8211 support for SDTS rasters and vectors, and OGR projections support.

I am currently inclined to merge the GDALDataset with OGRDataset, merge
driver registration mechanisms, and make it one library with uniform "GDAL"
naming. The configuration mechanism should include flags to just build in
raster, or vector support if desired, but it would be in the same vein as
just selecting a subset of formats. This merge would be a substantial job,
and would cause substantial disruption for users of OGR stuff, notable 
Safe Software. 

I am interested in opinions on this, though ultimately I am likely to make
the decision based on my own predilections. This problem has an erie
similarity to the discussion of whether to merge "TVP" new style vector
support back into GeoGateway many years ago at PCI. In the end there we
decided that having a unified model for accessing all sorts of geospatial
data look precidence over separability.

Finally, I would like to thank Atlantis Scientific who is currently supporting
my work on GDAL. I hope to be able to explain more of my work for Atlantis
in the not to distant future. 

Best regards,

---------------------------------------+--------------------------------------
I set the clouds in motion - turn up | Frank Warmerdam, warmerda at h...
light and sound - activate the windows | http://members.home.com/warmerda
and watch the world go round - Rush | Geospatial Programmer for Rent