OK, so Which bsd and which driver?

[gregory.menke at gsfc.nasa.gov]

Ralf Corsepius writes:
 > Am Mit, 2003-03-26 um 18.16 schrieb gregory.menke at gsfc.nasa.gov:
 > > Joking aside, Joel, please don't take this the wrong way- but this
 > > kind of thing is EXTREMELY frustrating.  I really wish these licensing
 > > issues were made more plain. 
 > You are mixing up things, here: 
 > Basically, RTEMS comes with no restrictions on licensing.
 > 
 > Linux is GPL'ed.
 > 
 > => Nothing prevents _you_ from using Linux-code with RTEMS if obeying
 > the restrictions of the GPL.
 > 
 > It's only that GPL'ed code can not be merged back into the "official
 > RTEMS sources" because this would imply having to put RTEMS under the
 > GPL.

Of course- thats my point.  I <want> to put my work back into RTEMS,
without imposing any sort of constraints on RTEMS.  I am quite aware
of the license issue- and am only annoyed because the incompatiblity
is not made public in any obvious way ahead of time.

 > 
 > >  A recursive grep for GPL in the rtems
 > > source tree didn't turn up any general warnings about inclusion of GPL
 > > code in RTEMS- only a couple ancient boilerplate references here and
 > > there.  Instead of wasting 2 days, I could have easily wasted a couple
 > > weeks getting the Linux 3com driver ported, working and efficient- and
 > > then have to throw it out due to the AFAICT essentially undocumented
 > > no-GPL requirement.  
 > As said above, it is not RTEMS which imposes restrictions on you, it is
 > Linux which does.

Correct.  And it would be nice if RTEMS contained a warning about this
issue.  Linux supports a huge variety of hardware these days and there
is and will be pressure to take advantage of that.  I realize it is
not strictly REQUIRED of RTEMS to make such warnings- however doing so
would be a polite act towards individuals contributing their labor to
the RTEMS community.

 > 
 > > Sure, I could use it in-house, which I probably
 > > would, but thats not helping the RTEMS community any.
 > 
 > Well, the idea of setting up a GPL'ed RTEMS and a non-GPL'ed one has
 > been discussed several times before, but AFAICT hasn't made it, so far
 > ....

I'm not advocating such a thing.  I don't get religious about software
licenses as such- the GPL and the bsd model serve somewhat different
constituencies, they both have advantages and I personally support
them both, morally and financially.  However, my grumpiness is due to
the fact that I would not be able to contribute a GPL derived driver
back into RTEMS- and that policy is not clearly stated ahead of time.

So, I'm dumping the work I did and will work off one or another bsd's
driver, which I'll run to ground at some point.  I'm looking at
freebsd's if_xl.c at the moment, which looks like it supports a
variety of 3com devices.

Gregm