[GRASSLIST:1367] Re: creating a desktop GIS application using GRASS

[Glynn Clements]

Radim Blazek wrote:

> > But I just feel that data I/O should maybe be treated differently. The
> > whole vector library should not be re-licenced as in 5.7 it includes
> > interfaces to advanced analysis functionality, e.g. directed graph stuff
> > and all the 'cleaning and tidying' of non-topological data (wasn't it
> > recently merged into the Vlib?). So there would need to be a separate
> > clearly defined I/O library (for reading&writing to the GRASS database) if
> > this re-licencing was going to happen. But Glynn already said he would
> > re-write the raster i/o functions for 5.7 so there is some scope there.
> 
> Better I/O than nothing. I don't quite understand, why do you fear,
> to relicense whole lib/vector/Vlib to LGPL? It is mostly original code 
> from CERL (public domain) + my improvements, so no big problem 
> with contributions under GPL. 
> Or do you worry that LGPL discourages new contributors?

It doesn't encourage new contributors to make their contributions
available to us. Someone writing proprietary extensions could use
LGPL'd code without providing anything in return.

Consider the situation where an individual or organisation extends
GRASS primarily for their own purposes (and, AFAICT, this is quite
common). Once they have a useful tool, what do they do with it? 
Depending upon the GRASS licensing conditions, they have either two or
three possible options:

1. Keep it to themselves.
2. Give it away (i.e. distribute it under the GPL).
3. Sell it (i.e. distribute it under terms which require per-copy
royalties).

Option 3 is only available if we allow it. Only option 2 benefits us. 
If we allow option 3, it may reduce the likelyhood of them choosing
option 2 (I'm fairly sure that it won't increase it).

Personally, I'm not even sure if using the LGPL for basic I/O is a
good idea. The only argument in favour of doing so with which I
sympathise is that it would reduce the likelyhood of a proprietary
product providing "partial" support for GRASS maps; that might result
in pressure on us to restrict ourselves to a limited "compatible"
subset.

More generally, the only area where I think that it's a good idea for
Free Software developers to actively support the developement of
proprietary software is in support of open data formats (zlib and PNG
are good examples). Although, even there, there are cases when it may
be better to impose more conditions in the hope of extracting more
benefit.

-- 
Glynn Clements <glynn.clements at virgin.net>