--opJtzjQTFsWo+cga
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

On Sat, Aug 26, 2006 at 11:25:59AM +0200, Timo Schoeler wrote:
> thus Bill Studenmund spake:
> >On Sat, Aug 26, 2006 at 12:49:03AM +0200, Timo Schoeler wrote:
> >>thus Mark Weinem spake:
> >>>But we should rnot support petitions and ask vendors for OS-specific
> >>>binary-only drivers. Nvdia drivers for FreeBSD and NetBSD without the=
=20
> >>>possibility for
> >>>all the other and smaller Open Source projects to profit - no thanks! =
=20
> >>>It's no problem for me that blob drivers work with Free- oder NetBSD  =
-=20
> >>>but we should really not demand them.
> >>yeah, and that's the point. as soon as one accepts a blob from any=20
> >>vendor they will say in future that you accepted it and don't see the=
=20
> >>necessity of opening their documentation (i.e. NOT the design itself,=
=20
> >>but the docs).
> >
> >I think you're wrong here. Well, you may be right about what is said, but
> >that's not what's important. My experience is that if someone says that
> >they will only blob a driver and that's ok because you took a blob
> >earlier, they are just finding an excuse. The decision to open-source or=
=20
> >not happened earlier.
>=20
> i think that you (still) didn't get the point; it's not about what's=20
> right or wrong today, or about which compromise to go.

You're wrong about what I think. You have elaborated the point that I
understood from you before. :-)

> it's more a psychological thing; (unfortunely) we're living in a=20
> *capitalistic* world. if vendor A doesn't have to publish documentation=
=20
> in order to get free (really free) operating systems support the card or=
=20
> whatever the vendor released, vendor B, C and D, who used to open up=20
> their documentation, will stop doing so.
>=20
> why? because there are managers and lawyers without *any* clue of what's=
=20
> going on in the real world -- they just want to optimize their company=20
> to make more money (profit), in most cases regardless of who has to die=
=20
> or whatever. in this business, usually nobody has to die, except openness.

I think you are ignoring some aspects of the market here.

Your arguement above assumes that there is no market pressure to open=20
source, which is why vendors B, C, and D would stop.

If, however, vendors B, C, and D saw revenue or positive intangibles (say=
=20
open video drivers led to SlickTasticGaming III running on Linux which led=
=20
to sales of SlickTasticGaming III which was optimizaed for vendor B's=20
graphics, which helped vendor B stay the "hot" vendor for GPUs) from open=
=20
sourced drivers, then they'd keep open sourcing.

> vendors B, C and D would save costs on documentation (for people outside=
=20
> their engineers), distribution of it, and most of all, feel more secure=
=20
> about their engineering, which they don't have to publish for any longer.

If the company gains nothing from releasing open source drivers, why=20
should it continue?

This gets us to what I think is wrong with the arguement you're making. I=
=20
think the arguements for open source driver support should be not that a=20
company "should" do it, but that a company will be able to make more $$=20
than the cost of open sourcing.

=46rom my experiences with the clueless managers and lawyers you mention
above, you're more likely to get nothing than even a blob driver. I don't
think that these actions will weaken our position as our position is most
likely already weak; we lost the arguement(s) before we got to this point. =
=20
It's like discussing how thick we should make a house's walls when we're
building the foundation on wet sand.

Instead if we explain how an open source driver can widen sales potential
(more than the cost of support), then we'll get better results. If we make
it easy to create open source drivers (documentation, documentation, code
suites, more documentation, etc.), we lower the cost for drivers. That
makes it easier to get companies to make them.

We also can be sneaky. The big problem I see with blob drivers isn't that
they're blobs, it's that they usually only come for i386. If as part of
developing our DDK we make it standard to build the driver for a bunch of
architectures, then it's easier to get vendors who make a driver for
NetBSD to make one for a lot of architectures.

> psychological, nothing else. it's like the child given a finger -- it=20
> demands the whole hand (it's a saying in germany, don't know if there's=
=20
> something equal in the UK/US).

There are similar expressions, but they involve reference to giving an=20
inch (2.5 cm) then getting asked for a yard (~ 1 m) or a mile (1.6 km).

To use that terminology, my thought is that chances are the child wanted
the hand before you even walked up, so giving a finger or not doesn't
matter.

Note: "to give someone the finger" is slang in the US for making a rude=20
hand jesture involving the middle finger.

Take care,

Bill

--opJtzjQTFsWo+cga
Content-Type: application/pgp-signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.3 (NetBSD)

iD8DBQFE8LJ1Wz+3JHUci9cRApOsAJ0WwNH3cDzXXCx0l2Pm4BdeOE6j7QCeLJu7
ui+8BseTwv68D5FdxH1Ne3E=
=bz7Y
-----END PGP SIGNATURE-----

--opJtzjQTFsWo+cga--