My 1 cent (after the exchange rate) on the subject of licensing and some
random unstructured ideas:
[ Note to the reader:
I won't feel bad if you hit delete and move on at this point]
My general thoughts are that relicensing 5.0/5.3 is impossible, 5.7
vector libraries maybe possible but not easy, and new 5.7+ LGPL raster
libraries should be possible, but it would need a new LGPL libgis2 for
things it talked to. And things libgis2 talked to would need to be
rewritten as well. It quickly becomes start the entire GIS from scratch.
[Do I have that right? Can a LGPL library not link to historical GPL?]
I am not too opposed to the LGPL/dual license for new stuff (it's the
authors' choice), but existing stuff is very messy.
Personally I prefer the GPL, but as I'm not the author of any of the
libraries it legally isn't my decision, I can just post to the mailing
list with my opinion like anyone else. If the authors of the libraries
want LGPL, that's their choice. I'm personally somewhat reluctant to
claim authorship over much more than large parts of d.legend (which will
be GPL). I'd also mention that I come from an academic background and a
lot of my support for the GPL comes from being sickened by the
corporatization of Universities all over the world. It's killing the
free exchange of scientific knowledge and Human growth. But that's
another rant. I also worked for a private engineering company for a
few years and am all too aware of other for-profits actively and
underhandedly trying to profit off of our hard work, so I'm a bit
sensitive to that angle as well.
Many projects (and their developers) run for a few years, then they
disappear. That's the nature of things and that's ok, when my current
project is done I expect to pick up and do something completely
different too. What the GPL gives us is the opportunity for new people
to pick up the pieces a year later and get it going again. If it wasn't
GPL, it would all disappear & be a constant stream of people beginning
over from 0 and only getting to working-beta before moving on. We're
standing on the shoulders of giants and all that. The GPL is a path to
lasting cumulative growth which will stick around, it's a one way system
which only gets better with time. Sorry if that's a bit trite, but I
think it is accurate.
A LGPL file I/O library might be a place to start, if there was a
_GRASS file format specification_ from which a new LGPL library/program
could be written from scratch. For the ascii site/old vector/raster
formats this could be done in a few minutes, maybe the "published" 5.0
Programmer's manual is enough already. Slow, I know. But it would be
something. Call it the "Worldwide Enviro-terrestrial Engineering
Database System" or WEEDS ;). It could use another common mapping
format, just make sure that GRASS totally supports it and the PROJ_INFO
specifications. GDAL & PROJ are MIT licensed, so they could act as the
go-between.. Maybe "gdal --with-grass=libgrass" is that already.
It would have to exist totally outside grass5, and not use any of the C
API; but I don't see why it couldn't be run from the bash prompt while
GRASS is still running, even if that's a bit cheeky. As an analogy, I'm
thinking of proprietary Linux software that uses the standard GNU tools
in their install scripts.
If the map format is open, any program can be written to do stuff to
that data and write back to that format. Nothing is stopping people from
writing software that will work along side GRASS and have output that is
readable by GRASS.
What I would not like to see is all the "value" & power of the new
vector libraries given away and GRASS see nothing in return. I think it
will be a very valuable tool for some amazing new models once people
start using it more (by the number of compile-help questions for 5.7
lately I think this is already happening). It would really help
accelerate the acceptance of 5.7 if we had such a killer-app model. *We*
won't get that model anytime soon by not being GPL, IMO. The GPL is the
price that is paid in exchange for the use of the vector/gis engine. I
don't think we'd (I'd) have any of the cool r.sun or r.terraflow models
to play with without the GPL. Being able to present my work in the
context of the rest of the ecological system with these 'freebee' extra
models has been really great. It's my "payment" for working on GRASS,
and there's little point in doing anything other than bug fixes for
myself if it isn't GPL, as I have no desire to be an unpaid employee of
someone elses for-profit venture.
I do think we have a responsibility to the previous developers be
careful not to do anything that might legally jeopardize the GRASS
copyright, if that's even possible.
Short answer: GPL, but I'm not opposed to people writing programs that
do really cool things with maps which were created with GRASS and that
GRASS can read the results of. That's their right and the way forward I
think.
Hamish