[GRASS5] libgrass license?

[Hamish]

> > Also, out of curiosity, what overlap is there between GRASS++ and
> > libgrass (which is LGPL) ?
> 
> GRASS++ itself is MIT, but it is using GRASS libraries also.

.. which means it can only be practically used as GPL until it no longer
links against GPL libraries (as non-GPL libraries become available), at
which point it magically reverts to MIT? Is that correct?


> I am not sure about libgrass, but it cannot be anything more free than
> GPL, as it is based on GRASS 5

From:
http://gdal.velocet.ca/projects/grass/

libgrass5-1.0.0.tar.gz/COPYING
"GNU LESSER GENERAL PUBLIC LICENSE 
Version 2.1, February 1999"

libgrass5-1.0.0.tar.gz/README
"The libgrass package consists of the majority of the GRASS libgis, and
libdatetime library build as a standalone shared library suitable for
use by non-GRASS applications wishing to read and write GRASS
databases."

One would have to check that the included "majority of the GRASS libgis,
and libdatetime" is only old public domain code and not newer GPL
additions, which cannot be arbitrarily re-licensed by a 3rd party as
LGPL, AFAIK.


Surely the file format isn't that complicated that GRASS libraries have
to be used for read/write. The programmer's manual should provide enough
info for a clean non-GPL library to be written, even if it is a pain to
reproduce working code. (which would be the GRASS++ library, which would
eventually be destined to be included with GDAL/OGR under the MIT/X
license?)


Currently GDAL skirts the issue by not including libgrass.
They should be careful about what options their official binaries
include though.

gdal/frmts/grass/frmt_grass.html:
"GDAL optionally supports reading of existing GRASS raster layers
(cells, and image groups), but not writing or export. The support for
GRASS raster layers is determined when the library is configured, and
requires libgrass to be pre-installed."



?,
Hamish