RE: [future of drivers?] a proposal for binary drivers.

RE: [future of drivers?] a proposal for binary drivers.

[Tim Tassonis]

> On Mon, 2006-03-13 at 10:24 +0100, Bernd Petrovitsch wrote:
>> Maybe not under (US-American) copyright, but under continental
>> European authors rights there are such possibilities (I leave it to
>> lawyers if and when they apply) and you can't even remove them with
>> contracts (as with all law stuff).
>> 
> 
> Please refrain from knee-jerk USA-bashing, you said yourself you don't
> know the law.
> 
> Lee
> 

Are you completely out of your head or what? When somebody reminds you 
of the fact that U.S. law is not the only law in the world and that some 
things may be different in other countries you call him a knee-jerking 
USA-basher???

In this sense, you better go back wanking to pictures of tortured 
iranians, yankee.

Tim

RE: [future of drivers?] a proposal for binary drivers.

[Lee Revell]

On Mon, 2006-03-13 at 20:00 +0100, Tim Tassonis wrote:
> > On Mon, 2006-03-13 at 10:24 +0100, Bernd Petrovitsch wrote:
> >> Maybe not under (US-American) copyright, but under continental
> >> European authors rights there are such possibilities (I leave it to
> >> lawyers if and when they apply) and you can't even remove them with
> >> contracts (as with all law stuff).
> >> 
> > 
> > Please refrain from knee-jerk USA-bashing, you said yourself you don't
> > know the law.
> > 
> > Lee
> > 
> 
> Are you completely out of your head or what? When somebody reminds you 
> of the fact that U.S. law is not the only law in the world and that some 
> things may be different in other countries you call him a knee-jerking 
> USA-basher???
> 

No I was referring to the implication that the GPL is not enforceable
under US copyright law.

> In this sense, you better go back wanking to pictures of tortured 
> iranians, yankee.

No comment

Re: [future of drivers?] a proposal for binary drivers.

[Tim Tassonis]

Lee Revell wrote:
> On Mon, 2006-03-13 at 20:00 +0100, Tim Tassonis wrote:
>>> On Mon, 2006-03-13 at 10:24 +0100, Bernd Petrovitsch wrote:
>>>> Maybe not under (US-American) copyright, but under continental
>>>> European authors rights there are such possibilities (I leave it to
>>>> lawyers if and when they apply) and you can't even remove them with
>>>> contracts (as with all law stuff).
>>>>
>>> Please refrain from knee-jerk USA-bashing, you said yourself you don't
>>> know the law.
>>>
>>> Lee
>>>
>> Are you completely out of your head or what? When somebody reminds you 
>> of the fact that U.S. law is not the only law in the world and that some 
>> things may be different in other countries you call him a knee-jerking 
>> USA-basher???
>>
> 
> No I was referring to the implication that the GPL is not enforceable
> under US copyright law.

Well, I'd really like to know how your patriotic sentiments were hurt by 
Bernd in the above sentence. Maybe we Europeans are not compassionate 
enough, it seems, as I can see absolutely nothing anti-american in his 
posting. Maybe mentioning the existence of other countries/laws is 
nowadays considered anti-american?

Tim

Re: [future of drivers?] a proposal for binary drivers.

[Lee Revell]

On Mon, 2006-03-13 at 20:25 +0100, Tim Tassonis wrote:
> Lee Revell wrote:
> > On Mon, 2006-03-13 at 20:00 +0100, Tim Tassonis wrote:
> >>> On Mon, 2006-03-13 at 10:24 +0100, Bernd Petrovitsch wrote:
> >>>> Maybe not under (US-American) copyright, but under continental
> >>>> European authors rights there are such possibilities (I leave it to
> >>>> lawyers if and when they apply) and you can't even remove them with
> >>>> contracts (as with all law stuff).
> >>>>
> >>> Please refrain from knee-jerk USA-bashing, you said yourself you don't
> >>> know the law.
> >>>
> >>> Lee
> >>>
> >> Are you completely out of your head or what? When somebody reminds you 
> >> of the fact that U.S. law is not the only law in the world and that some 
> >> things may be different in other countries you call him a knee-jerking 
> >> USA-basher???
> >>
> > 
> > No I was referring to the implication that the GPL is not enforceable
> > under US copyright law.
> 
> Well, I'd really like to know how your patriotic sentiments were hurt by 
> Bernd in the above sentence. Maybe we Europeans are not compassionate 
> enough, it seems, as I can see absolutely nothing anti-american in his 
> posting. Maybe mentioning the existence of other countries/laws is 
> nowadays considered anti-american?

I'm sorry I posted that, this thread is OT anyway.  I'm just annoyed
when people who don't know US law speculate about how terrible it is
(like the common misconception that reverse engineering for
interoperability is banned in the US).  Of course I don't think that
pointing out differences is anti-American.

Lee

Re: [future of drivers?] a proposal for binary drivers.

[Tim Tassonis]

>>> under US copyright law.
>> Well, I'd really like to know how your patriotic sentiments were hurt by 
>> Bernd in the above sentence. Maybe we Europeans are not compassionate 
>> enough, it seems, as I can see absolutely nothing anti-american in his 
>> posting. Maybe mentioning the existence of other countries/laws is 
>> nowadays considered anti-american?
> 
> I'm sorry I posted that, this thread is OT anyway.  I'm just annoyed
> when people who don't know US law speculate about how terrible it is
> (like the common misconception that reverse engineering for
> interoperability is banned in the US).  Of course I don't think that
> pointing out differences is anti-American.
> 
> Lee

And I'm sorry for my language (well, I'm actually not, but of course I 
only wanted to provoke). The thing is probably that Non-Americans often 
get a bit irritated when U.S. posters start talking about law on lklm 
and not seldom seem to imply that the U.S. law is the only relevant one 
in regard of Linux. Of course the GPL was written in the U.S., but Linux 
was GPL'ed back in Finland, Linus used to be non-american, Alan Cox 
still is and the number of Linux installation outside the U.S. certainly 
far outnumbers the ones within the states. And when the pointing out of 
this fact is considered anti-american I get quite furious. At least we 
Europeans have not much else in the computing industry apart from Linux, 
you know (apart from having invented the computer).

Tim

RE: [future of drivers?] a proposal for binary drivers.

[Bernd Petrovitsch]

On Mon, 2006-03-13 at 14:15 -0500, Lee Revell wrote:
[...]
> No I was referring to the implication that the GPL is not enforceable
> under US copyright law.

Sorry, but where and how did I my sentences implies that?

The email spoke about "libraries" and "applications" which have under
the GPL also legal semantics (especially if you use both in the same
sentence) which IMHO do not apply to "kernel and drivers". Or the
original author sees the core kernel as a "library" and his driver as an
"application" (which is IMHO in the sense of th reasoning if I read the
analogy correct).

And the whole chain of reasoning lacks IMHO important conditions like
under which rules you actually acquired the rights to the source you
*based your work on*. If your work is not based on that GPL source, it
is a completely other thing (and discussion).

So at most I killed the prove/reasoning as such. But this doesn't imply
that it's result is wrong (as there may be another one).

If the GPL would not be enforcable under US law (which is AFAIK not the
case), the GPL would be probably dead anyway since it was written in the
US by US people with the US jurisdiction in mind AFAIK.

You should hear Austrian lawyers speaking about their problems with "the
GPL and Austrian law" (or German law - if Austria is too small to be
relevant - which is not that different in many aspects).
Yes, it got better in the last year.


	Bernd
-- 
Firmix Software GmbH                   http://www.firmix.at/
mobil: +43 664 4416156                 fax: +43 1 7890849-55
          Embedded Linux Development and Services

Re: [future of drivers?] a proposal for binary drivers.

[Matt Reuther]

Hi, David.

You seem to have misspelled, "I am not a lawyer. Please seek the advice of a 
competent attorney experienced in copyright law in your jurisdiction instead 
of listening to my uninformed blather."

Hope this helps!

Matt

Re: [future of drivers?] a proposal for binary drivers.

[Bernd Petrovitsch]

On Mon, 2006-03-13 at 01:02 -0500, Matt Reuther wrote:
> Hi, David.
> 
> You seem to have misspelled, "I am not a lawyer. Please seek the advice of a 
> competent attorney experienced in copyright law in your jurisdiction instead 
> of listening to my uninformed blather."
> 
> Hope this helps!

Not at all since you're mail misses References: headers to get threading
accurately to work.

	Bernd
-- 
Firmix Software GmbH                   http://www.firmix.at/
mobil: +43 664 4416156                 fax: +43 1 7890849-55
          Embedded Linux Development and Services

RE: [future of drivers?] a proposal for binary drivers.

[David Schwartz]

> Hi, David.
>
> You seem to have misspelled, "I am not a lawyer. Please seek the
> advice of a
> competent attorney experienced in copyright law in your
> jurisdiction instead
> of listening to my uninformed blather."
>

	I will thank you not to put words in my mouth. If you wish to express your
own opinion, and even your own opinion of my opinion, you may do so. But
implying that your opinions are my own is out of line. Even rude.

	DS

RE: [future of drivers?] a proposal for binary drivers.

[Lee Revell]

On Mon, 2006-03-13 at 14:00 -0800, David Schwartz wrote:
> implying that your opinions are my own is out of line. Even rude.
> 

Yes, it is rude - as are long OT threads about copyright law and the GPL
on this mailing list.

Lee

RE: [future of drivers?] a proposal for binary drivers.

[David Schwartz]

> On Mon, 2006-03-13 at 14:00 -0800, David Schwartz wrote:
> > implying that your opinions are my own is out of line. Even rude.

> Yes, it is rude - as are long OT threads about copyright law and the GPL
> on this mailing list.

	Even assuming that, there is a huge difference between a personally rude
action taken by one person against another person and a rude result that
arises from the uncoordinated actions of many people. Are you trying to
imply that I somehow deserve to have people be personally rude to me?

	DS

Re: [future of drivers?] a proposal for binary drivers.

[Tim Tassonis]

 >
 > if there was binary allowed (with any license) maybe dlink themself
 > would build a driver, make documentation and provide it on CD, just
 > see how much effort would be saved and in end he would get more time
 > to treat his patients.
 >

Apart from all the other good arguments already posted:

Are you really sure they will? Maybe dlink will, but I can tell from 
personal expierience (whine, whine) that the majority of vendors still 
won't release drivers. For the simple reason because they regard Linux a 
market too small to support. That is the main reason for most of them, 
not the license stuff.

Before Linux, I was an OS/2 user and although every vendor in the world 
was allowed to provide OS/2 drivers, there were more or less the same 
amount of vendor contributed drivers as there are now in Linux.

I'm 100% sure that Linux supports more hardware than OS/2 did back then 
and the user base (Desktop wise) was at least as big as Linux.

OS/2 died exactly because software companies didn't write closed-source 
software, hardware companies didn't write closed-source drivers, and IBM 
couldn't write it all themselves.

So why repeat this desaster?
Tim

Re: [future of drivers?] a proposal for binary drivers.

[Randy.Dunlap]

On Wed, 8 Mar 2006, Tim Tassonis wrote:

>  >
>  > if there was binary allowed (with any license) maybe dlink themself
>  > would build a driver, make documentation and provide it on CD, just
>  > see how much effort would be saved and in end he would get more time
>  > to treat his patients.
>  >
>
> Apart from all the other good arguments already posted:
>
> Are you really sure they will? Maybe dlink will, but I can tell from
> personal expierience (whine, whine) that the majority of vendors still
> won't release drivers. For the simple reason because they regard Linux a
> market too small to support. That is the main reason for most of them,
> not the license stuff.

Right, what we really need IMO is specs and the right to produce
GPL drivers from the specs.  Very little real work is required from the
vendors aside from IP/legal.

> Before Linux, I was an OS/2 user and although every vendor in the world
> was allowed to provide OS/2 drivers, there were more or less the same
> amount of vendor contributed drivers as there are now in Linux.
>
> I'm 100% sure that Linux supports more hardware than OS/2 did back then
> and the user base (Desktop wise) was at least as big as Linux.
>
> OS/2 died exactly because software companies didn't write closed-source
> software, hardware companies didn't write closed-source drivers, and IBM
> couldn't write it all themselves.
>
> So why repeat this desaster?
> Tim

-- 
~Randy

Re: [future of drivers?] a proposal for binary drivers.

[Jan Knutar]

On Wednesday 08 March 2006 23:28, Tim Tassonis wrote:

> OS/2 died exactly because software companies didn't write closed-source 
> software, hardware companies didn't write closed-source drivers, and IBM 
> couldn't write it all themselves.

I read somewhere that the development kit was obscenely expensive and
effectively locked out all hobbyists and small companies.

Re: [future of drivers?] a proposal for binary drivers.

[Helge Hafting]

Jan Knutar wrote:

>On Wednesday 08 March 2006 23:28, Tim Tassonis wrote:
>
>  
>
>>OS/2 died exactly because software companies didn't write closed-source 
>>software, hardware companies didn't write closed-source drivers, and IBM 
>>couldn't write it all themselves.
>>    
>>
>
>I read somewhere that the development kit was obscenely expensive and
>effectively locked out all hobbyists and small companies.
>
IBM stuff may have been expensive.
There was gcc for os/2 though, which was nice for hobbyists.
The way I remember it, drivers wasn't the big problem.  The lack of drivers
could have become a problem if the lack of software was solved.  But it
wasn't - no need to drive a device for which you don't have an app . . .

Having gcc meant one could port some gpl software, which usually
originated on linux or other unixes.  But using such software on
linux was always easier, and so I switched os.

Helge Hafting

Re: [future of drivers?] a proposal for binary drivers.

[Anshuman Gholap]

I knew this was sore issue, i knew some people would say to me, i am
just another troll barking at tree not knowing anything , and i knew
loads of people will share same views or atleast do have lots of
questions regarding EULA, LAW, COPYRIGHT, "does it suit fundamental's
of linux", and similar things to say but boy, i never never never
expected this many views going back and forth, some real live
developer issue someone mentioned here (glucometer) did give some
sense of how_it_gets difficult for a developer to_do kernel related
things. due to my work i wasnt able to reply to this post, just
reading all replies here was overwhelming (ofcourse gmail did made it
lots easier).

(think of this i shud had put my hosting company name at start, that
wud had been cheap but plenty of adverting :) but i wanted and still
want the_issue to be core of this thread)

Linus might either be thinking "boy is this the longest and oldest
ever rant or what" or "this needs some serious attending, gotta talk
about this with morton" or, "this pisses me off to maximum,  mustttt
resists pulling my eyes out sticking fingers in and swirling my brain
around" .


I agree asking for  modular/binary driver option in linux kernel is
trollling, as asking for open source windows to billyg . but why did i
ask? no i didnt wake up one day and felt i shud bark at some tree,
this is same kind of trolling like , if someone with old pc saying how
sad it is for him with no updates available for win98 and slowing apps
and no modern applications suck for him and  i in reply to give a
suggestion like "have you tried linux? its pretty neat now, give it a
try".

 i already mentioned at starting of thread what drove me to writing
this topic and i think linus should seriously give a thought or review
to this.

I even read a long time some japanese company (maybe mobile company
docomo, i really dont know the name) were asking to make binary driver
legal and more integrated with linux kernel to which linus vehemently
opposed it, i googled "binary drivers" and found loads of article like
in 1999 and similar where linus has shown his displeasure. but who
knows, times change, needs change, people too change.

To end this email, I from heart want to thank and congratulate to all
developers users coders and others who shared , challenged,
confronted, discussed their views, it really educated me as it must
have done to lot others too.

Thank you.

Re: [future of drivers?] a proposal for binary drivers.

[Dave Neuer]

On 3/13/06, Anshuman Gholap <anshu.pg@gmail.com> wrote:
>
>  i already mentioned at starting of thread what drove me to writing
> this topic and i think linus should seriously give a thought or review
> to this.

It is not his decision, he is just one (albeit an important,
influential one) of something like 900 copyright holders in the
kernel.

[future of drivers?] a proposal for binary drivers.

[Anshuman Gholap]

Hello,

First, I would like to give you my intro, so that you have a picture
who is talking. I am 28 year old from india working as administrator
for a hosting company (not naming here cause that will be cheap way to
advert it).  We have 100+ servers (shared,semi-dedicated and
dedicated) and growing steadily, all have linux-distros ofcourse (many
RHEL).

But i also have 3-4 pc's with me and one laptop, most of them have 
linux installed, i go digicam not working on linux, webcam not working
on linux, dlink 630d pcmcia card not working natively on linux, the
list is just too big here (sony 510i not working, etc).

I had read earlier Linus Travolds totally opposes the idea of allowing
binary drivers to exists along with kernel, but on other hand, i also
heard lots like, how andrew morton is stretched for supporting the
programmers at various companies who keep sending him drivers for
their devices .

1) The people who have control over linux kernel as of now are
countable on hands but cannot be held countable for work cause they do
it as hobby/"insert anything which says working for free", now for a
peice of code like linux kernel, such kind of aloofness regarding
manpower and kind_of nazism in not allowing others to dynamically
get_work_done (like binary driver) seems totally wrong.

2) there are two possibilities here, a) linus and co can gather in a
building pay all intellects and allow fast driver developments, b)
allow binary drivers to work with linux kernel dynamically(with their
own license what they choose).

b) ofcourse is like china accepting democracy cause that the only way
to continue living, but although it sounds that extreme, i can see
ONLY THAT to happen sooner or later when one day linus is not part of
the team controlling linux kernel, so why not start to make it happen
right now and shape it the way it can be benificial to everyone?
like there is mm kernel we can have kernel-dri-2.*** which the desktop
users can use knowingly that third party drivers can work with mixture
of lincenses. there even can be rating system for a company which can
be rated for their quality of drivers, so the users know before hand.
how good or bad the driver (even if he/she isnt codeguru).

This email is very raw and not polished at all, this is just a bunch
of thoughts which has came to my head after working with linux based
distros on home computers as well as enterprise servers years and
years now.

If something i said here is wrong , i apologise and correct me where
you feel so.

Regards,
Anshuman Gholap
admin.

Re: [future of drivers?] a proposal for binary drivers.

[Jan Knutar]

On Wednesday 08 March 2006 11:35, Anshuman Gholap wrote:

> linux installed, i go digicam not working on linux, webcam not working

I thought cameras in general did usb masstorage thing and thus
worked with anything?

> countable on hands but cannot be held countable for work cause they do
> it as hobby/"insert anything which says working for free", now for a

Actually quite alot of them do it for work, including Torvalds and Cox.

> peice of code like linux kernel, such kind of aloofness regarding
> manpower and kind_of nazism in not allowing others to dynamically
> get_work_done (like binary driver) seems totally wrong.

Unfortunately the license (GPL) was chosen along time ago and can't
really be changed at this point.
 
> 2) there are two possibilities here, a) linus and co can gather in a
> building pay all intellects and allow fast driver developments, b)
> allow binary drivers to work with linux kernel dynamically(with their
> own license what they choose).

The real question is: Why do binary-only drivers need to exist?


> b) ofcourse is like china accepting democracy cause that the only way
> to continue living, but although it sounds that extreme, i can see
> ONLY THAT to happen sooner or later when one day linus is not part of
> the team controlling linux kernel, so why not start to make it happen
> right now and shape it the way it can be benificial to everyone?

GPL.

You could try direct your efforts at *BSD which has a more liberal
licensing policy.
Microsoft, for example, has a version of .NET for FreeBSD, but not
for Linux. Presumably because of the license differences.

> like there is mm kernel we can have kernel-dri-2.*** which the desktop
> users can use knowingly that third party drivers can work with mixture
> of lincenses. there even can be rating system for a company which can
> be rated for their quality of drivers, so the users know before hand.

Already in place. The "Tainted" flag.

In general: Binary drivers == bad.
After that, there are several levels of bad...

Besides... Looking at Win64, it looks like hardware manufacturers
have problems coming up with drivers... If they had done them
open-source for Linux or any other opensource operating system
in the first place, they'd most likely have less problems with the 64
bit transition right now.... Binary-only just hurts everyone.

> This email is very raw and not polished at all, this is just a bunch
> of thoughts which has came to my head 

The same here.

Re: [future of drivers?] a proposal for binary drivers.

[Anshuman Gholap]

Hello Jan,

you said  quote, "The real question is: Why do binary-only drivers
need to exist?"

super super super nice question.

ok here is the deal, My bro who is a doctor and has lenovo laptop,
buys lets say dlink pcmcia wifi card , and opens the box, gets the
hardware out and the software  cd out, all he sees is windows related
drivers and documentation, he and any person like him wont even bother
how to plug this in ubuntu linux (which i almost mind-controlled him
into installing it) , he knowing me as a linux person will keep
bugging me, when i tell him to install a kernel source compile it to
allow 16k stack, install ndiswrapper and load the windows driver and
compile install gtk-wifi app and get wifi network.  he might admit me
into hospital for talk_while_geek with a normal person.

if there was binary allowed (with any license) maybe dlink themself
would build a driver, make documentation and provide it on CD, just
see how much effort would be saved and in end he would get more time
to treat his patients.

I have thousands of similar scenarios. Even I wont mind the luxury of
making hardware just working and not going to google>>download src>>if
bug/error found>>go to forums post thread>>hang on irc and bug
ppl>>get more things compiled done >>if work then enjoy>> or wait for
the philanthropic coder to solve bug and release new ver.

Best regards,
Anshuman Gholap.

Re: [future of drivers?] a proposal for binary drivers.

[Martin Mares]

Hello!

> if there was binary allowed (with any license) maybe dlink themself
> would build a driver, make documentation and provide it on CD,

Maybe. But what prevents them to releasing the driver if binaries
are not allowed?

> I have thousands of similar scenarios. Even I wont mind the luxury of
> making hardware just working and not going to google>>download src>>if
> bug/error found>>go to forums post thread>>hang on irc and bug
> ppl>>get more things compiled done >>if work then enjoy>> or wait for
> the philanthropic coder to solve bug and release new ver.

"Binary drivers will make all devices just work" is a dream. Maybe a nice
one, but just a dream. It seldom works this way in Windows (almost every
time you run into some minor, but annoying bug noone is willing to fix),
why should it work with Linux?

				Have a nice fortnight
-- 
Martin `MJ' Mares   <mj@ucw.cz>   http://atrey.karlin.mff.cuni.cz/~mj/
Faculty of Math and Physics, Charles University, Prague, Czech Rep., Earth
The number of UNIX installations has grown to 10, with more expected.  (6/72)

Re: [future of drivers?] a proposal for binary drivers.

[Jan Engelhardt]

[-- Attachment #1: Type: TEXT/PLAIN, Size: 695 bytes --]

>
>"Binary drivers will make all devices just work" is a dream. Maybe a nice
>one, but just a dream. It seldom works this way in Windows (almost every
>time you run into some minor, but annoying bug noone is willing to fix),
>why should it work with Linux?
>

It's (should be: it was) worse than that. Under Win98 (Common Platform For
Binary Drivers), you had to install a usbdisk driver for every usb product
class, i.e. Samsung only provides a mass storage driver for Samsung, SONY only
for SONY drives, etc. Luckily this is a little relieved in Win2000 (CPFBD too)
where there is a generic usbdisk driver à la usb_storage.ko, which reads 
more than just from one vendor.


Jan Engelhardt
-- 

Re: [future of drivers?] a proposal for binary drivers.

[Jan Knutar]

On Wednesday 08 March 2006 12:03, you wrote:

> allow 16k stack, install ndiswrapper and load the windows driver and
> compile install gtk-wifi app and get wifi network.  he might admit me
> into hospital for talk_while_geek with a normal person.

Yeah... and if manufacturers stopped being stubborn, and just made
it possible to create open-source drivers, all your friend would have
to do would be install the distro of his choice and it would all "just
work"... As it does currently with most ethernet cards. It just works.
Why? Because the drivers are opensource and in the kernel.

> I have thousands of similar scenarios. Even I wont mind the luxury of
> making hardware just working and not going to google>>download src>>if
> bug/error found>>go to forums post thread>>hang on irc and bug
> ppl>>get more things compiled done >>if work then enjoy>> or wait for
> the philanthropic coder to solve bug and release new ver.

I feel the pain too. I used to do this.

Eventually the binary drivers stepped on eachother's toes and exploded
spectaculary, and there was no fix to be found. I just gave up, and am now
using computer with mostly opensource drivers... Only item I haven't been
able to get away from so far is the dialup modem, which I have to use
binary drivers for... But I only use it in emergencies anyway, since the cost-free
drivers are only able to do 14.4kbit.

Actually, I'd be happy if the binary drivers actually MANAGED to get a 14.4K
connection... Anything above 2400bps seems futile. It's not the telephone line,
either, my old 33.6K modem worked reliably at 33600 constantly. Then thunder
struck and fried it :-(

The conexant softmodem was the only one available in the local store, and I
recalled there were binary drivers for it. Pity they don't work that well :(
The softmodem cost only 50E. A "real" modem is about 150E -250E or
something, and I don't want to spend that kind of money on something
that only gives me 56kbit...

I use my cellphone these days... It has horrendous latency, but atleast it
talks standard Hayes AT through its dataport.

Re: [future of drivers?] a proposal for binary drivers.

[Diego Calleja]

El Wed, 08 Mar 2006 13:11:42 +0200,
Jan Knutar <jk-lkml@sci.fi> escribió:


> The softmodem cost only 50E. A "real" modem is about 150E -250E or
> something, and I don't want to spend that kind of money on something
> that only gives me 56kbit...

A real 56 kbit modem using standard COM ports costs arount 30-40 €
in spain.

Re: [future of drivers?] a proposal for binary drivers.

[Luke-Jr]

On Wednesday 08 March 2006 11:11, Jan Knutar wrote:
> :( The softmodem cost only 50E. A "real" modem is about 150E -250E or
> something, and I don't want to spend that kind of money on something that
> only gives me 56kbit...

Fund a project to implement a software modem. While a "real" modem might have 
better performance, softmodems are more of a raw interface which can be 
better in the long run-- for example, if some new super modulation is 
produced for 1mbit over regular phone lines, you could possibly just upgrade 
your modem software for the new feature. You can also use the modem for voice 
capabilities, and have your computer act as an answering machine while you're 
not using it for the internet.
(Note this is all theoretically possible, and might require actual coding to 
achieve; just pointing out that softmodem isn't necessarilly worse than 
hardmodems)

Re: [future of drivers?] a proposal for binary drivers.

[Phillip Susi]

Luke-Jr wrote:
> Fund a project to implement a software modem. While a "real" modem might have 
> better performance, softmodems are more of a raw interface which can be 
> better in the long run-- for example, if some new super modulation is 
> produced for 1mbit over regular phone lines, you could possibly just upgrade 
> your modem software for the new feature. You can also use the modem for voice

Except that is not going to happen because there are laws of physics 
that must be obeyed.  The phone line is encoded with 64 Kbps inside the 
digital phone network so there is no possible way to modulate more data 
than that.  There is also the problem that the A/D converter ( which is 
all the "software modem" is -- it's basically a stripped down sound card 
) only runs fast enough to support 56,000 bps.  Even if it could run 
faster, that would place even _more_ load on the cpu.


> capabilities, and have your computer act as an answering machine while you're 
> not using it for the internet.
> (Note this is all theoretically possible, and might require actual coding to 
> achieve; just pointing out that softmodem isn't necessarilly worse than 
> hardmodems)

They ARE necessarily worse than real modems in that they require a fair 
amount of cpu cycles to perform all the DSP.  IIRC, this also has to be 
done in interrupt context to maintain the low latency required, which 
lowers the interactive responsiveness of the system - even when you are 
not transferring much data.

Re: [future of drivers?] a proposal for binary drivers.

[Michael Concannon]

Luke-Jr wrote:
> On Wednesday 08 March 2006 11:11, Jan Knutar wrote:
>   
>> :( The softmodem cost only 50E. A "real" modem is about 150E -250E or
>>     
Is "E" Euros?  Is so, really? 56K modems look to be about $10-20(USD) 
around here (US NE)

> Fund a project to implement a software modem. While a "real" modem might have 
> better performance, softmodems are more of a raw interface which can be 
> better in the long run-- for example, if some new super modulation is 
> produced for 1mbit over regular phone lines, you could possibly just upgrade 
> your modem software for the new feature. You can also use the modem for voice 
> capabilities, and have your computer act as an answering machine while you're 
> not using it for the internet.
>   
This is a disease... "future proofing" is a lie to get you to buy 
hardware/software that you don't need... by the time you are ready to 
use your cell phone as an ground penetrating radar treasure hunting 
device, you will likely be ready to toss the old one anyway because the 
battery only lasts 5minutes and the power cord on the charger that costs 
3X the phone is frayed.

I am not saying that there are not things which are really are 
upgradeable (word processors, image manipulators, media players, 
compilers etc...), I am saying that in the long run when you consider 
unit cost, man power, electricity consumption, heat dissipation, there 
are a number of tasks being offloaded into software that are far less 
"efficient" and cost effective on a net basis.

Using your PC as an POTS answering machine is a waste of electricity... 
(of course that assumes that you are not running a  multi-gigatexel-per 
second OpenGL screen saver when you are not around in which case all is 
lost, go ahead)... If you must, why waste GHz cycles to do what a KHz 
embedded DSP can do for $10 (retail)?

We have a long way to go before PCs are anywhere close to as efficient 
as they can be about power consumption, but it won't help if we all buy 
TVs and toasters with quad core 64bit processors because we might want 
to start our own personal SETI project ... someday...

When you consider the cost of a device, consider the whole cost of it:
a. the time required to install maintain with each new Fedora/Ubuntu 
release every 8 minutes...
b. the cost of not being able to use it because the softmodem only works 
on this or that rev of OS/Motherboard, etc...
c. the cost of "exploits" in your softmodem stack allowing dial up 
hacking...
d. silly regulations which force binary blobs to keep me from roasting 
my neighbor with 802.11 cranked up to ... well "11"...
e. the cost of the power consumed doing DSP in a CPU rather than a 
dedicated circuit

There are applications where it will be hard to tell and applications 
where fast paced change in technology means upgrading every few months 
is a reality.  In those cases software content makes sense...

Then there modems which have changed VERY little 10+ years... 

Sorry for the rant - in my business I am watching company after company 
die trying to finish the bloated and impossibly complex software for 
hardware that has been sitting in the lab for a year... perhaps if the 
line between hardware and software were drawn a little more 
realistically,  that would not happen...   It is frustrating to watch...

Software is not free, no matter how little it costs to obtain it, it 
costs a LOT... The "disease" is "just do it in software", the cure is 
actually designing a solution and figuring out where software makes 
sense... Every part of my PC that has drastically increased its software 
content has meant more binary drivers followed by lots of hacking and 
then finally open source support - right around the time I am about to 
replace it with the next shiny new thing...

I am continually dumbfounded that the "driver" layer of PCs is as 
complex as it is... I know why which is that hardware companies cannot 
be troubled to actually finish designing things before they sell them, 
so the driver layer gives them a nice gasket layer to fix all the bugs...

I'll stop ranting now...

/mike

> (Note this is all theoretically possible, and might require actual coding to 
> achieve; just pointing out that softmodem isn't necessarilly worse than 
> hardmodems)
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
>
>   

Re: [future of drivers?] a proposal for binary drivers.

[Helge Hafting]

Anshuman Gholap wrote:

>Hello Jan,
>
>you said  quote, "The real question is: Why do binary-only drivers
>need to exist?"
>
>super super super nice question.
>
>ok here is the deal, My bro who is a doctor and has lenovo laptop,
>buys lets say dlink pcmcia wifi card , and opens the box, gets the
>hardware out and the software  cd out, all he sees is windows related
>drivers and documentation, he and any person like him wont even bother
>how to plug this in ubuntu linux (which i almost mind-controlled him
>into installing it) , he knowing me as a linux person will keep
>bugging me, when i tell him to install a kernel source compile it to
>allow 16k stack, install ndiswrapper and load the windows driver and
>compile install gtk-wifi app and get wifi network.  he might admit me
>into hospital for talk_while_geek with a normal person.
>
>if there was binary allowed (with any license) maybe dlink themself
>would build a driver, make documentation and provide it on CD, just
>see how much effort would be saved and in end he would get more time
>to treat his patients.
>  
>
This is where your assumption is _WRONG_. 

Sure - it would be nice if dlink provided a driver!

But there is _no need_ for this driver to be binary!

You see, providing a open-source driver is no harder than
a binary one for the provider.  This because _all_ drivers have
source.  If the open driver is for somewhat popular hardware then
the company won't even have to maintain it - someone will
usually do it.  So usually, a company can save money by
providing source.

Second, an open source driver will make life much much easier
for the end-users too.  Much easier, in fact, than a binary driver
provided on a cd.  Why?  Because you actually have to install
binary drivers.  That's work an end-user don't want to do.

An open-source driver however, will be integrated into the
linux kernel source.  That means that when you install linux
(or have someone else install it, as "computer dummies" do,)
then the driver is in the machine already. 

When a linux distribution is on your machine, then all the
open source drivers supported by linux is there already.
You can add hardware, and it will work either immediately
(USB, cardbus, other hotplug stuff) or after the next boot
(PCI, IDE, other non-hotplug stuff.)

Of course, you have to take care to buy hardware that
linux actually support.  Not a big problem, for linux
support all categories of hardware even if there certainly
is unsupported hardware too.

This is no different from the windows world - you don't
want to buy unsupported hardware there either.
And yes, such hardware exist. The obvious example
is any computer with an intel-incompatible processor,
many of which runs linux just fine. :-)

>I have thousands of similar scenarios. Even I wont mind the luxury of
>making hardware just working and not going to google>>download src>>if
>bug/error found>>go to forums post thread>>hang on irc and bug
>ppl>>get more things compiled done >>if work then enjoy>> or wait for
>the philanthropic coder to solve bug and release new ver.
>  
>
Linux surely have the luxury of hardware that just works.
Poor windows users - they actually need to mess around
with the driver CDs that comes with most hardware. I don't!
And when they reinstall, they need all those CDs again. I don't.
And if they stumble onto a bug - and yes, they do sometimes,
then they can write to microsoft and only hope to
get any answer at all. When was the last time microsoft (or
someone providing windows drivers for their hardware)
fixed a bug and released an update a few _hours_ after
a bug report?  With linux, that happens all the time,
although there sure are bugs that take longer too.


To put it short - the problem you describes is easily
solved by drivers.  There is no reason whatsoever
for them to be binary though.  (As an end-user, your
brother won't need the compiler, he'll get the driver
precompiled from his linux distribution vendor - _if_
they have access to an open source driver.)

Helge Hafting

Re: [future of drivers?] a proposal for binary drivers.

[Matthias Andree]

On Wed, 08 Mar 2006, Anshuman Gholap wrote:

> if there was binary allowed (with any license) maybe dlink themself
> would build a driver, make documentation and provide it on CD, just
> see how much effort would be saved and in end he would get more time
> to treat his patients.

And you assume doctor would care to compile the kernel himself in his
working hours, rather than ask the NOC staff of $hospital to do that for
him?  You expect Linux would ship with these binary drivers for any
device?  All major Linux distributions (including Turbo, Red Hat and
Fedora, Novell, SUSE and Opensuse, Debian, Ubuntu, Gentoo) could work
out a single way of installing drivers?  Interesting.

> I have thousands of similar scenarios. Even I wont mind the luxury of
> making hardware just working and not going to google>>download src>>if
> bug/error found>>go to forums post thread>>hang on irc and bug
> ppl>>get more things compiled done >>if work then enjoy>> or wait for
> the philanthropic coder to solve bug and release new ver.

That coder needs to be provided with full specs, and that is what
usually does not happen, because then the hardware isn't as easily
exchangable to save 0.003 US$ per device, many vendors fear their
"intellectual property" (as though thoughts had property rights, brains
have...).

-- 
Matthias Andree

Re: [future of drivers?] a proposal for binary drivers.

[Bernd Petrovitsch]

On Wed, 2006-03-08 at 15:33 +0530, Anshuman Gholap wrote:
[...]
> into installing it) , he knowing me as a linux person will keep
> bugging me, when i tell him to install a kernel source compile it to
> allow 16k stack, install ndiswrapper and load the windows driver and

And you seriously think that $COMPANY will rewrite their driver to work
with 4K-stacks (which seems to me to be an absolute requirement ATM)?

[...]
> if there was binary allowed (with any license) maybe dlink themself
> would build a driver, make documentation and provide it on CD, just

Here are too many "maybe"s and "would" in there. Do you have a written
contract or similar stuff?

And who is maintaining that driver and solving all possibly related
problems?
It is not fun to debug software where some unknown piece of code may
have introduced a bug and you can't chase it down since you don't have
the source.
Will $COMPANY have enough capable people to follow LKML and look after
bug reports involving their driver as long as the driver should be
considered maintained?
Will $COMPANY provide versions for not-Intel CPUs where people may put
their hardware into?

The trivial solution is: Don't buy that hardware if you want to run it
on Linux.

Alas, the big difference is: In the Windows world, the hardware
companies are interested to solve problems with their drivers (otherwise
they have no business), in the Linux world they are not.
And that won't change with "officially" allowing binary drivers.

The consequence would be that $COMPANY writes a driver and blames the
rest of the Linux world to change some internal undocumented interface
months lateron just that they can commercially state to "support Linux"
but without any real reason. In the non-evolutionary Windows world this
holds until the next major release, but not on the high-tech front.

	Bernd
-- 
Firmix Software GmbH                   http://www.firmix.at/
mobil: +43 664 4416156                 fax: +43 1 7890849-55
          Embedded Linux Development and Services

RE: [future of drivers?] a proposal for binary drivers.

[David Schwartz]

> The consequence would be that $COMPANY writes a driver and blames the
> rest of the Linux world to change some internal undocumented interface
> months lateron just that they can commercially state to "support Linux"
> but without any real reason. In the non-evolutionary Windows world this
> holds until the next major release, but not on the high-tech front.

	It should be possible to define a reasonable set of requirements that one
must meet in order to claim to "support Linux", just as Microsoft does for
Windows. One requirement should definitely be that driver source code be
available to everyone who purchases the hardware and that all (or at least
sufficient) interfaces to the hardware be well-documented.

	DS

Re: [future of drivers?] a proposal for binary drivers.

[Valdis.Kletnieks]

[-- Attachment #1: Type: text/plain, Size: 927 bytes --]

On Tue, 14 Mar 2006 00:06:48 +0100, Bernd Petrovitsch said:
> On Wed, 2006-03-08 at 15:33 +0530, Anshuman Gholap wrote:
> [...]
> > into installing it) , he knowing me as a linux person will keep
> > bugging me, when i tell him to install a kernel source compile it to
> > allow 16k stack, install ndiswrapper and load the windows driver and
> 
> And you seriously think that $COMPANY will rewrite their driver to work
> with 4K-stacks (which seems to me to be an absolute requirement ATM)?

>From the NVidia drivers changelog:

2004-6-30 version 1.0-6106

    * Added support for GLSL (OpenGL Shading Language).

    * Added support for GL_EXT_pixel_buffer_object.

    * Added support for 4kstack kernels.

Looks like they managed to do that quite some time ago - in fact, before
some parts of the *in-kernel* code were totally cleaned up....

So yes, I *do* expect $COMPANY to re-write their driver to support 4K stacks. ;)

[-- Attachment #2: Type: application/pgp-signature, Size: 228 bytes --]

Re: [future of drivers?] a proposal for binary drivers.

[Bernd Petrovitsch]

On Tue, 2006-03-14 at 19:50 -0500, Valdis.Kletnieks@vt.edu wrote:
> On Tue, 14 Mar 2006 00:06:48 +0100, Bernd Petrovitsch said:
> > On Wed, 2006-03-08 at 15:33 +0530, Anshuman Gholap wrote:
> > [...]
> > > into installing it) , he knowing me as a linux person will keep
> > > bugging me, when i tell him to install a kernel source compile it to
> > > allow 16k stack, install ndiswrapper and load the windows driver and
> > 
> > And you seriously think that $COMPANY will rewrite their driver to work
> > with 4K-stacks (which seems to me to be an absolute requirement ATM)?
> 
> From the NVidia drivers changelog:

NVidia is one of the better examples (and I leave the binary driver
discusion out) - they supported their drivers from the start (and the
first years there were lots of trouble with official builds every other
day or so IIRC).

[...]
> Looks like they managed to do that quite some time ago - in fact, before
> some parts of the *in-kernel* code were totally cleaned up....
> 
> So yes, I *do* expect $COMPANY to re-write their driver to support 4K stacks. ;)

Of course implies "maintaing a driver for Linux" that such
maintenance/development/call-it-what-you-want is done (and not only for
4K-stacks - this just a current example and probably needs handling on
the driver side and providing some "compatibility layer" won't work that
good).

My doubt is that (above supposed old-economy) $COMPANY (which was more
or less "forced" to support Linux and didn't "freely" choose that way
like NVidia) writes a driver (or payed someone external once for it) and
considers the "Linux case" closed for the next 3 years (as you would
with a Win*-driver).

	Bernd
-- 
Firmix Software GmbH                   http://www.firmix.at/
mobil: +43 664 4416156                 fax: +43 1 7890849-55
          Embedded Linux Development and Services

Re: [future of drivers?] a proposal for binary drivers.

[Martin Michlmayr]

* Jan Knutar <jk-lkml@sci.fi> [2006-03-08 11:51]:
> > linux installed, i go digicam not working on linux, webcam not
> I thought cameras in general did usb masstorage thing and thus
> worked with anything?

It's fairly common for new cameras, but it's definitely not the case
in general.  There's a program called gphoto2 that can talk to a
number of cameras that don't use USB mass storage, though.
-- 
Martin Michlmayr
tbm@cyrius.com

Re: [future of drivers?] a proposal for binary drivers.

[Xavier Bestel]

On Wed, 2006-03-08 at 10:35, Anshuman Gholap wrote:
> If something i said here is wrong , i apologise and correct me where
> you feel so.

I know you posted in good faith, but you are in fact part of a periodic
troll. Please search the mailing-lists archives where this subject has
been "debated" several times a year, you should understand why this is a
Bad Idea.

Thanks,

	Xav

Re: [future of drivers?] a proposal for binary drivers.

[Al Viro]

On Wed, Mar 08, 2006 at 03:05:11PM +0530, Anshuman Gholap wrote:
> 1) The people who have control over linux kernel as of now are
> countable on hands but cannot be held countable for work cause they do
> it as hobby/"insert anything which says working for free", now for a
> peice of code like linux kernel, such kind of aloofness regarding
> manpower and kind_of nazism in not allowing others to dynamically
> get_work_done (like binary driver) seems totally wrong.
 
> This email is very raw and not polished at all

Indeed - self-godwinated right in the opening salvo.  Tsk, tsk...  Trolls
these days...

Re: [future of drivers?] a proposal for binary drivers.

[Anshuman Gholap]

well ya, I knew i was running the risk to be labelled like that, cause
i thought to talk of this issue, more shake is needed that just stir.

please dont get me wrong (even though i think most of you already
have), i own my graditude for the livelihood i am having to
linux,linus and co.

this discussion is totally for betterment of the new users who should
not be forced to become developers in order to get a trival thing
running on their desktop/laptop, like device driver.

Now that i think of this, a more better person to speak on this would
be device manufacturer's themselves, is it the state_of_current linux
prohibiting them from developing drivers? do they hate oss so much?
would their mentality and willingness change if binary/mixed license
drivers were allowed to work with kernel.

if you can send this thread link to dlink or any other manufacturer, please do.

for now, i am going to shut my yapper.

Anshuman.

On 3/8/06, Al Viro <viro@ftp.linux.org.uk> wrote:
> On Wed, Mar 08, 2006 at 03:05:11PM +0530, Anshuman Gholap wrote:
> > 1) The people who have control over linux kernel as of now are
> > countable on hands but cannot be held countable for work cause they do
> > it as hobby/"insert anything which says working for free", now for a
> > peice of code like linux kernel, such kind of aloofness regarding
> > manpower and kind_of nazism in not allowing others to dynamically
> > get_work_done (like binary driver) seems totally wrong.
>
> > This email is very raw and not polished at all
>
> Indeed - self-godwinated right in the opening salvo.  Tsk, tsk...  Trolls
> these days...
>

Re: [future of drivers?] a proposal for binary drivers.

[Arjan van de Ven]

> this discussion is totally for betterment of the new users who should
> not be forced to become developers in order to get a trival thing
> running on their desktop/laptop, like device driver.

so you only want to take, but not give. 
Then you're posting to the wrong mailing list.

Re: [future of drivers?] a proposal for binary drivers.

[Anshuman Gholap]

well, apart from sony 510i cell, and cam, i have managed to get
installed everything that i need for my work, or which i cant live
without,

I was speaking of all this, cause i am tired to everyone peskering me,
arguing with me how linux cannot work with my "insert new hardware" ,
I have to google stuff out and hand-hold them to get it working, this
was cool to do some years ago, when i didnt bother to work full time
and felt good helping others, now that feel good factor has ran out,
and i get more annoyed than anything else. I dont think i can explain
why did i post this. maybe someone somewhere will get it.

anyway, nevermind.

sorry to anyone who was hurt/offended here. sorry if linus reads this
and gets pissed off, sorry to cox and morton if they feel same.

peace,
anshuman.
On 3/8/06, Arjan van de Ven <arjan@infradead.org> wrote:
>
> > this discussion is totally for betterment of the new users who should
> > not be forced to become developers in order to get a trival thing
> > running on their desktop/laptop, like device driver.
>
> so you only want to take, but not give.
> Then you're posting to the wrong mailing list.
>
>
>

Re: [future of drivers?] a proposal for binary drivers.

[Jon Masters]

On 3/8/06, Anshuman Gholap <anshu.pg@gmail.com> wrote:

> I was speaking of all this, cause i am tired to everyone peskering me,
> arguing with me how linux cannot work with my "insert new hardware" ,
> I have to google stuff out and hand-hold them to get it working, this
> was cool to do some years ago, when i didnt bother to work full time
> and felt good helping others, now that feel good factor has ran out,
> and i get more annoyed than anything else. I dont think i can explain
> why did i post this. maybe someone somewhere will get it.

It strikes me that this thread isn't really about binary-only drivers
and that whole ugly discussion. What it's really about is getting
third party drivers to work with Linux more easily. That conversation
does involve the "out of tree vs. upstream" question, but mention of
binary-only drivers actually only serves to cloud the real issue here.

Jon.

Re: [future of drivers?] a proposal for binary drivers.

[sean]

On Tue, 14 Mar 2006 03:02:12 +0000
"Jon Masters" <jonmasters@gmail.com> wrote:

> It strikes me that this thread isn't really about binary-only drivers
> and that whole ugly discussion. What it's really about is getting
> third party drivers to work with Linux more easily. That conversation
> does involve the "out of tree vs. upstream" question, but mention of
> binary-only drivers actually only serves to cloud the real issue here.

But that debate has been settled.   The only possible solution to making 
3rd party drivers easier to use is to make a stable driver-ABI.  And the
reason that isn't going to happen has been explained many times already.

Linux is designed around having all drivers in the source tree, which 
removes the problems experienced by out-of-tree drivers.  The pain felt
by those promoting and using 3rd party drivers is brought on _themselves_
by being unable or unwilling to adapt to the Linux environment.

Demanding that Linux adapt to the needs of 3rd party drivers ignores 
the fact that this will just shift the pain onto the core kernel
developers.   That pain would ultimately be felt by all Linux users 
who would then get a kernel which is artifically constrained by all 
these old interfaces.  It would destroy one of the key advantages open
source offerings have over the closed source.   In other words, it
would do more harm than good.

In the end, most of the demands for Linux to "adapt" come from people 
who really don't appreciate or embrace open source.   I think kernel
developers are protecting Linux (and thus its users) by ignoring these
often vitriolic demands for it to conform to the way things are done
in the closed source world.

Cheers,
Sean

Re: [future of drivers?] a proposal for binary drivers.

[Matthias Schniedermeyer]

Anshuman Gholap wrote:
> well ya, I knew i was running the risk to be labelled like that, cause
> i thought to talk of this issue, more shake is needed that just stir.
> 
> please dont get me wrong (even though i think most of you already
> have), i own my graditude for the livelihood i am having to
> linux,linus and co.

To get to the point of the others binary-only-discussions.

You only see that you can't use a device today.
I know that is annoying, but you have to see the "big picture":

Less hostility regarding binary-only drivers would lead to a "flood" of
binary-only-drivers which are undebuggable and unmaintanable by the kernel
developers. IOW you would be at the mercy of the vendor of the device to
make a compatible driver in the future.

But there is a planet-size catch:
Vendors think in money. So if you have a device that is end of line most
vendors couldn't care less if you can't use it anymore with current systems.
Given that the vendor is still in business after all!

So instead of having a paper-weight today you will have it a few years later.
I don't see the big difference.

IOW. A "new" device may be working today, but will be a paper-weight
later.
Whereas an "old" device will be a paper-weight today, if the vendor only
provided binary-only drivers "back then" when it was "new".

In contrast most times you have an OSS-driver it will work "indefinetly",
as it can be maintained over the years.

It's all a shifting of who is hurt and when. In the long run the current
model should be working better and better. Whereas binary-only drivers
would destroy/undermine the achievements we have now.




-- 
Real Programmers consider "what you see is what you get" to be just as
bad a concept in Text Editors as it is in women. No, the Real Programmer
wants a "you asked for it, you got it" text editor -- complicated,
cryptic, powerful, unforgiving, dangerous.

Re: [future of drivers?] a proposal for binary drivers.

[Anshuman Gholap]

thanks mathhais, this made huge sense to me.

I will soon going to get a laptop ( centrino duo models are looking
great, but kindof out_of_my budget). I will see the vendors who  dont
have native linux drivers are not in that model, as a non-developer,
this is the only thing in_my_hand.

i have no idea why did i make the horrible decision to get dlink
router and pcmcia card, where as the linksys set has a working (some
ppl complain) native driver support for its cards. wish i can turn the
clock back.

again, thank you mattias for putting it into nice prespective :)

Regards,
Anshu.
On 3/8/06, Matthias Schniedermeyer <ms@citd.de> wrote:
> Anshuman Gholap wrote:
> > well ya, I knew i was running the risk to be labelled like that, cause
> > i thought to talk of this issue, more shake is needed that just stir.
> >
> > please dont get me wrong (even though i think most of you already
> > have), i own my graditude for the livelihood i am having to
> > linux,linus and co.
>
> To get to the point of the others binary-only-discussions.
>
> You only see that you can't use a device today.
> I know that is annoying, but you have to see the "big picture":
>
> Less hostility regarding binary-only drivers would lead to a "flood" of
> binary-only-drivers which are undebuggable and unmaintanable by the kernel
> developers. IOW you would be at the mercy of the vendor of the device to
> make a compatible driver in the future.
>
> But there is a planet-size catch:
> Vendors think in money. So if you have a device that is end of line most
> vendors couldn't care less if you can't use it anymore with current systems.
> Given that the vendor is still in business after all!
>
> So instead of having a paper-weight today you will have it a few years later.
> I don't see the big difference.
>
> IOW. A "new" device may be working today, but will be a paper-weight
> later.
> Whereas an "old" device will be a paper-weight today, if the vendor only
> provided binary-only drivers "back then" when it was "new".
>
> In contrast most times you have an OSS-driver it will work "indefinetly",
> as it can be maintained over the years.
>
> It's all a shifting of who is hurt and when. In the long run the current
> model should be working better and better. Whereas binary-only drivers
> would destroy/undermine the achievements we have now.
>
>
>
>
> --
> Real Programmers consider "what you see is what you get" to be just as
> bad a concept in Text Editors as it is in women. No, the Real Programmer
> wants a "you asked for it, you got it" text editor -- complicated,
> cryptic, powerful, unforgiving, dangerous.
>
>

Re: [future of drivers?] a proposal for binary drivers.

[Rik van Riel]

On Wed, 8 Mar 2006, Matthias Schniedermeyer wrote:

> But there is a planet-size catch:
> Vendors think in money. So if you have a device that is end of line most
> vendors couldn't care less if you can't use it anymore with current systems.
> Given that the vendor is still in business after all!

It's worse than that, actually.

Your old device no longer working is not a neutral thing
for the manufacturer, but actually a GOOD thing, since
you will be forced to buy a new device - one that has a
driver for a new version of the OS.

Binary only drivers are a good way to ensure that devices
become obsolete sooner than they would otherwise.  Not so
good for users, though...

-- 
All Rights Reversed

Re: [future of drivers?] a proposal for binary drivers.

[Dave Neuer]

On 3/8/06, Anshuman Gholap <anshu.pg@gmail.com> wrote:
>
> this discussion is totally for betterment of the new users who should
> not be forced to become developers in order to get a trival thing
> running on their desktop/laptop, like device driver.

So far, this discussion is not helping new users at all.

Users need to do a little research before buing hardware, period.
Things are getting better across all platforms and operating systems
(remember the bad old days of jumpers and IRQ conflicts?), it is not
linux-specific.

What is linux-specific in this context is that many people, like
myself, who have contributed code to the kernel under the GPL *don't
want* their code to be used in non-free software, period. Someone who
wants to leverage my work needs to do it under the terms that I allow.
That is the law. Whining is not going to change my mind.

If a company thinks they can make money selling hardware with
closed-source drivers (on some other OS), more power to them. If a
company thinks they can make money selling hardware with open-source
drivers on Linux and want to leverage my work, more power to them
(I'll even help them). But they can't use my work and not release the
code.

It's really that simple.


Dave

Re: [future of drivers?] a proposal for binary drivers.

[Hannu Savolainen]

On Wed, 8 Mar 2006, Dave Neuer wrote:

> What is linux-specific in this context is that many people, like
> myself, who have contributed code to the kernel under the GPL *don't
> want* their code to be used in non-free software, period. Someone who
> wants to leverage my work needs to do it under the terms that I allow.
> That is the law. Whining is not going to change my mind.
> 
> If a company thinks they can make money selling hardware with
> closed-source drivers (on some other OS), more power to them. If a
> company thinks they can make money selling hardware with open-source
> drivers on Linux and want to leverage my work, more power to them
> (I'll even help them). But they can't use my work and not release the
> code.
You are mixing two different things. Binary driver is not the same thing 
as binary-only driver. Being binary means just that the driver is 
distributed as a precompiled module. However the driver may still be open 
sourced (under GPL or whatever license you like). The full source 
code may be shipped with the installable binary or be distributed in some 
other way for users who want to compile it themselves. 

In fact all Linux distributions ship binary drivers. The user can install 
the kernel/driver sources but most "ordinary" users don't install them. 

Best regards,

Hannu
-----
Hannu Savolainen (hannu@opensound.com)
http://www.opensound.com (Open Sound System (OSS))
http://www.compusonic.fi (Finnish OSS pages)
OH2GLH QTH: Karkkila, Finland LOC: KP20CM

RE: [future of drivers?] a proposal for binary drivers.

[David Schwartz]

Dave Neuer wrote:

> What is linux-specific in this context is that many people, like
> myself, who have contributed code to the kernel under the GPL *don't
> want* their code to be used in non-free software, period. Someone who
> wants to leverage my work needs to do it under the terms that I allow.
> That is the law. Whining is not going to change my mind.

	I'm sorry, that's not the law. You can wish it was, but it's not.

	If the law allowed you to give your software away for free and then put
restrictions on use, you could drop copies of a poem from an airplane (or
put it up on a billboard) and then demand royalties from everyone who read
it.

	DS

Re: [future of drivers?] a proposal for binary drivers.

[Dave Airlie]

>
> > What is linux-specific in this context is that many people, like
> > myself, who have contributed code to the kernel under the GPL *don't
> > want* their code to be used in non-free software, period. Someone who
> > wants to leverage my work needs to do it under the terms that I allow.
> > That is the law. Whining is not going to change my mind.
>
>         I'm sorry, that's not the law. You can wish it was, but it's not.
>
>         If the law allowed you to give your software away for free and then put
> restrictions on use, you could drop copies of a poem from an airplane (or
> put it up on a billboard) and then demand royalties from everyone who read
> it.

We don't give away our software for free, we give it away under the
terms of the GPL, you must abide by the terms of the GPL to use it.
that is the law, you can wish it wasn't but it is.

Dave.

Re: [future of drivers?] a proposal for binary drivers.

[Rudolf Randal]

I was wondering how many direct letters from lkml people have been
written to vendors asking for cooperation with regard to specs, joint
development or other solutions for open source drivers for their
devices.
If there has been letters of this kind (there might have been many)-
do they exist in the open? has there been replies?
I dont know how many dlink people (or other vendors) are reading lkml
or other linux related maillists/websites but I do know that the tone
of lkml often isnt pleasant when it comes to licensing issues, closed
vs open source (GPL). I too feel strongly about oss and GPL - but the
tone on this list (often) might be too much to handle for some people!
Maybe request for an open discussion with some of these vendors would
bring about some of their concerns over IP and other issues and could
maybe even open up for some progress ??

Rudolf
--
Rudolf Randal - Hässleholmsgatan 3B lgh 503 - 214 43 Malmö - Sweden -
Phone: +46 (0)76 234 05 77

Re: [future of drivers?] a proposal for binary drivers.

[DervishD]

    Hi Rudolf :)

 * Rudolf Randal <rudolf.randal@gmail.com> dixit:
> I was wondering how many direct letters from lkml people have been
> written to vendors asking for cooperation with regard to specs,
> joint development or other solutions for open source drivers for
> their devices.

    Not exactly kernel related, but anyway: I wrote almost a month
ago to Menarini Diagnostics. They make glucometers, and the latest
model they sell can be connected to the PC with a special cable. Of
course, Windows.

    Since no Linux support is provided for this useful device, I did
an offer to Menarini: if they provided me with the cable and the
specs, I'll write a Linux driver and/or a Linux application to
support *their* product at no cost. I would do it for free because
I'm very interested in connecting my glucometer to my Linux box.

    I made clear that the driver and/or app would be GPL'd, but on
the other hand I made clear too that I won't release the specs they
would give me if they didn't want to (except of course those bits
exposed by the source code itself).

    I haven't got any answer from them. I can use other glucometer,
of course, but this one is, IMHO, the best one in the market and I
don't think any other vendor would give me specs for their products.

    And please note that I'm not talking about all the specifications
of the glucometer: they can keep secret the way the glucometer makes
the measurements, I don't give a heck about that. I only want the
*communication* specs, so I'm able to retrieve data from the
glucometer through the cable (which is USB, AFAIK, so probably it
won't need a new driver).

> I too feel strongly about oss and GPL - but the tone on this list
> (often) might be too much to handle for some people!

    Probably, but if I make something for free, I think I should have
the right to impose distribution conditions, and GPL restrictions
seem to be very reasonable to me: if you use my work, you cannot
avoid others using it, and if you make a modification of my work, you
cannot distribute a binary of that modified work without the source.

    For me, not getting an answer from a vendor is too much to
handle. Having to buy a cable myself and doing reverse engineering to
make something I own work with my operating system is too much to
handle.

    I don't know how much Menarini paid for the Windows version of
the software. Probably more than the entire Linux kernel will cost to
produce in its entire life. The cost of a Linux version for them will
be a cable and a PDF document of the communication specification,
which probably will be a known protocol. They don't want to do it
because: a) they don't understand a word of the open source movement
and how it can be used to save money; or b) they're so stupid,
stubborn and money avid that won't do anything that doesn't involve
money, even if the money goes out of their own pockets!

> Maybe request for an open discussion with some of these vendors
> would bring about some of their concerns over IP and other issues
> and could maybe even open up for some progress ??

    I utterly agree. The tone in this list may be "strong", but it is
an open list nonetheless and I think all Linux hackers out there will
be happy to hear from some vendor doing a cooperation offer.

    Up to this point, Menarini doesn't seem to be one of them,
though.

    Raúl Núñez de Arenas Coronado

-- 
Linux Registered User 88736 | http://www.dervishd.net
http://www.pleyades.net & http://www.gotesdelluna.net
It's my PC and I'll cry if I want to... RAmen!

Re: [future of drivers?] a proposal for binary drivers.

[Andreas Mohr]

Hi,

On Thu, Mar 09, 2006 at 12:22:59PM +0100, DervishD wrote:
>     Not exactly kernel related, but anyway: I wrote almost a month
> ago to Menarini Diagnostics. They make glucometers, and the latest
> model they sell can be connected to the PC with a special cable. Of
> course, Windows.

[...]

>     I haven't got any answer from them. I can use other glucometer,
> of course, but this one is, IMHO, the best one in the market and I
> don't think any other vendor would give me specs for their products.
Possibly you wrote one mail only?
Then try again... they might easily be too busy or whatever.

>     And please note that I'm not talking about all the specifications
> of the glucometer: they can keep secret the way the glucometer makes
> the measurements, I don't give a heck about that. I only want the
> *communication* specs, so I'm able to retrieve data from the
> glucometer through the cable (which is USB, AFAIK, so probably it
> won't need a new driver).
Be certain to use libusb if possible instead of a new USB driver
(which you seem to indicate properly above)

>     I don't know how much Menarini paid for the Windows version of
> the software. Probably more than the entire Linux kernel will cost to
> produce in its entire life. The cost of a Linux version for them will
> be a cable and a PDF document of the communication specification,
> which probably will be a known protocol. They don't want to do it
> because: a) they don't understand a word of the open source movement
> and how it can be used to save money; or b) they're so stupid,
> stubborn and money avid that won't do anything that doesn't involve
> money, even if the money goes out of their own pockets!
You're talking strange things here.
"Probably more than the entire Linux kernel will cost to produce in its
entire life" - yah, right!
And you're jumping to conclusions and resorting to name calling
about their (non-)behaviour.

Andreas Mohr

Re: [future of drivers?] a proposal for binary drivers.

[Alistair John Strachan]

On Thursday 09 March 2006 11:22, DervishD wrote:
>     Hi Rudolf :)
>
>  * Rudolf Randal <rudolf.randal@gmail.com> dixit:
> > I was wondering how many direct letters from lkml people have been
> > written to vendors asking for cooperation with regard to specs,
> > joint development or other solutions for open source drivers for
> > their devices.
>
>     Not exactly kernel related, but anyway: I wrote almost a month
> ago to Menarini Diagnostics. They make glucometers, and the latest
> model they sell can be connected to the PC with a special cable. Of
> course, Windows.
>
>     Since no Linux support is provided for this useful device, I did
> an offer to Menarini: if they provided me with the cable and the
> specs, I'll write a Linux driver and/or a Linux application to
> support *their* product at no cost. I would do it for free because
> I'm very interested in connecting my glucometer to my Linux box.
>
>     I made clear that the driver and/or app would be GPL'd, but on
> the other hand I made clear too that I won't release the specs they
> would give me if they didn't want to (except of course those bits
> exposed by the source code itself).
>
>     I haven't got any answer from them. I can use other glucometer,
> of course, but this one is, IMHO, the best one in the market and I
> don't think any other vendor would give me specs for their products.

Sounds like reverse engineering territory. Fortunately there is software for 
windows that can help you reverse engineer USB or serial-over-USB protocols 
fairly easily. I doubt such a device would be difficult to write a driver 
for.

In my opinion, no answer isn't a good enough answer.

-- 
Cheers,
Alistair.

'No sense being pessimistic, it probably wouldn't work anyway.'
Third year Computer Science undergraduate.
1F2 55 South Clerk Street, Edinburgh, UK.

Re: [future of drivers?] a proposal for binary drivers.

[DervishD]

    Hi Alistair :)

 * Alistair John Strachan <s0348365@sms.ed.ac.uk> dixit:
> On Thursday 09 March 2006 11:22, DervishD wrote:
> >     I haven't got any answer from them. I can use other glucometer,
> > of course, but this one is, IMHO, the best one in the market and I
> > don't think any other vendor would give me specs for their products.
> 
> Sounds like reverse engineering territory. Fortunately there is
> software for windows that can help you reverse engineer USB or
> serial-over-USB protocols fairly easily. I doubt such a device
> would be difficult to write a driver for.

    Not difficult at all, once you build the cable (it's not a
standard USB cable), but I don't want to do reverse engineering this
time. Adding Linux support to this device will help the manufacturer
(probably a very tiny little help, but help nonetheless), who doesn't
seem to care enough of their users.

    I mean, the vendor is earning lots of money from me (not due to
the device, but to the supplies I need daily), so if they don't want
to collaborate (at no cost from them) to add Linux support, I'll just
use another glucometer that will give support. Here in Spain, all
glucometer-makers are more than happy to get my Menarini glucometer
and give me one of theirs at no charge (given that they will get the
money from the supplies). I can try almost all vendors. It's a pity,
because I think this meter is the best one in the market, but if I
have to use another product in order to have it supported by Linux,
I'll do, and I will advice others against the product I just
rejected.
 
> In my opinion, no answer isn't a good enough answer.

    I think the same. They may have very good reasons to not give the
specs, and I can understand them... if they tell me! Currently, I can
only make guesses about the reasons...

    Raúl Núñez de Arenas Coronado

-- 
Linux Registered User 88736 | http://www.dervishd.net
http://www.pleyades.net & http://www.gotesdelluna.net
It's my PC and I'll cry if I want to... RAmen!

RE: [future of drivers?] a proposal for binary drivers.

[David Schwartz]

> > > What is linux-specific in this context is that many people, like
> > > myself, who have contributed code to the kernel under the GPL *don't
> > > want* their code to be used in non-free software, period. Someone who
> > > wants to leverage my work needs to do it under the terms that I allow.
> > > That is the law. Whining is not going to change my mind.
> >
> >         I'm sorry, that's not the law. You can wish it was, but
> > it's not.
> >
> >         If the law allowed you to give your software away for
> > free and then put
> > restrictions on use, you could drop copies of a poem from an
> > airplane (or
> > put it up on a billboard) and then demand royalties from
> > everyone who read
> > it.

> We don't give away our software for free,

	Yes, you do. You can receive a copy of the Linux kernel without paying any
fee or agreeing to any license. You do in fact give away the software for
free.

> we give it away under the
> terms of the GPL,

	That is true. However, the GPL can only grant additional rights, it cannot
take any away.

> you must abide by the terms of the GPL to use it.

	No, that's not true. The law permits the normal expected use simply by
acquiring a copy of the work lawfully. Otherwise, you could buy a book and
then discover that you still had to pay a royalty to read it. Or you could
put a poem on a billboard and sue anyone who read it.

> that is the law, you can wish it wasn't but it is.

	You might try *reading* the law sometime.

	DS

Re: [future of drivers?] a proposal for binary drivers.

[Dagfinn Ilmari Mannsåker]

"David Schwartz" <davids@webmaster.com> writes:

> 	If the law allowed you to give your software away for free and
> then put restrictions on use, you could drop copies of a poem from an
> airplane (or put it up on a billboard) and then demand royalties from
> everyone who read it.

No. Copyright does not cover reading. It covers the distribution of
copies (and derived works) of works. You could just fine drop copies of
a poem from an airplain and then demand royalties from everoyone who
distributes additional copies of it (copies they made, not the copies
you dropped, cf. the doctrine of first sale).

-- 
ilmari

RE: [future of drivers?] a proposal for binary drivers.

[David Schwartz]

> "David Schwartz" <davids@webmaster.com> writes:

> > 	If the law allowed you to give your software away for free and
> > then put restrictions on use, you could drop copies of a poem from an
> > airplane (or put it up on a billboard) and then demand royalties from
> > everyone who read it.

> No. Copyright does not cover reading. It covers the distribution of
> copies (and derived works) of works. You could just fine drop copies of
> a poem from an airplain and then demand royalties from everoyone who
> distributes additional copies of it (copies they made, not the copies
> you dropped, cf. the doctrine of first sale).

	Why would you say "No" and then write something that 100% agrees with me?

	DS

Re: [future of drivers?] a proposal for binary drivers.

[Dave Neuer]

On 3/8/06, David Schwartz <davids@webmaster.com> wrote:
>
> Dave Neuer wrote:
>
> > What is linux-specific in this context is that many people, like
> > myself, who have contributed code to the kernel under the GPL *don't
> > want* their code to be used in non-free software, period. Someone who
> > wants to leverage my work needs to do it under the terms that I allow.
> > That is the law. Whining is not going to change my mind.
>
>        I'm sorry, that's not the law. You can wish it was, but it's not.
>
>        If the law allowed you to give your software away for free and then put
> restrictions on use, you could drop copies of a poem from an airplane (or
> put it up on a billboard) and then demand royalties from everyone who read
> it.

Right. Copyright law doesn't allow me to demand royalties from the
people who read my dropped-from-airplane poem. Copyright law *does*
say that I can prevent people from copying my dropped-from-airplane
poem and re-publishing it, however. Copyright law also absolutely
gives me monopoly power over derivitive works - you can't even create
a work derived from my dropped-from-airplane poem without my
permission, much less distribute one.

But let's not keep torturing the analogy. We're talking about
software, in particular driver software which is presumably (and here
a court gets to decide whether the presumption is true, not driver
authors) a derived work of the kernel on which it depends. Many, many
kernel developers have repeatedly stated their reservations of their
rights under copyright law and the GPL with regard to derived works on
this forum and elsewhere.

Best regards,
Dave

RE: [future of drivers?] a proposal for binary drivers.

[David Schwartz]

> Right. Copyright law doesn't allow me to demand royalties from the
> people who read my dropped-from-airplane poem.

	Right. In other words, simply by lawfully receiving the work, they have the
right to the normal, expected use of that work.

> Copyright law *does*
> say that I can prevent people from copying my dropped-from-airplane
> poem and re-publishing it, however.

	True, but this discussion wasn't about redistribution.

> Copyright law also absolutely
> gives me monopoly power over derivitive works - you can't even create
> a work derived from my dropped-from-airplane poem without my
> permission, much less distribute one.

	No, that's not true on two counts.

	First, your last part "much less distribute one" is utterly false.
Copyright law does not give you any special rights to restrict the
distribution of derivative works.

	Second, your first part, that it gives you monopoly power over the creation
of derivative works is also false. First sale and fair use can give people
the right to create derivative works.

> But let's not keep torturing the analogy. We're talking about
> software, in particular driver software which is presumably (and here
> a court gets to decide whether the presumption is true, not driver
> authors) a derived work of the kernel on which it depends. Many, many
> kernel developers have repeatedly stated their reservations of their
> rights under copyright law and the GPL with regard to derived works on
> this forum and elsewhere.

	If the driver is not a derived work of the kernel, then we agree you are
free to create and distribute it, I think. If the driver contains only as
much of the kernel as it must in order to be created at all, then under at
least United States precedent, it is not considered a derived work.

	Copyright only applies when there's more than one way to do the same thing.
If it is impossible to create a Linux kernel driver without taking X, then X
cannot be protected by copyright because it is practically necessary. (And
courts have never been impressed by arguments like "you don't need to create
a driver really" or "you can create a driver for another operating system"
because these are ways to express other ideas, not other ways to express the
same idea.)

	You cannot copyright an idea. "A Foo2000 SCSI driver for Linux 2.6" is an
*idea*. So you cannot argue that you have copyright on every practically
possible way to create such a driver.

	This argument could only apply in a case where the driver author took more
of the kernel than they practically had to in order to express the idea they
wished to express. (See Lexmak v. Static Controls, for example.)

	DS

Re: [future of drivers?] a proposal for binary drivers.

[Dave Neuer]

On 3/9/06, David Schwartz <davids@webmaster.com> wrote:
>
> > Copyright law also absolutely
> > gives me monopoly power over derivitive works - you can't even create
> > a work derived from my dropped-from-airplane poem without my
> > permission, much less distribute one.
>
>        No, that's not true on two counts.
>
>        First, your last part "much less distribute one" is utterly false.
> Copyright law does not give you any special rights to restrict the
> distribution of derivative works.

You're being rather hyper-technical and semantic here, aren't you? You
clearly can't distribute something you can't legally "prepare" in the
first place.

>
>        Second, your first part, that it gives you monopoly power over the creation
> of derivative works is also false. First sale and fair use can give people
> the right to create derivative works.

OK, 14 USC 106(2) gives me monopoly power, subject to certain
restrictions. You seemed to have missed my point, as none of the
restrictions you mentioned include creation of derivative software
programs for commercial distribution.

In another reply, you wrote:
>
>        You cannot copyright an idea. "A Foo2000 SCSI driver for Linux 2.6" is an
> *idea*. So you cannot argue that you have copyright on every practically
> possible way to create such a driver.

Your argument, if extended to fiction, is equivalent to "'An
elaboration of Gone With the Wind' is an idea." There may be such a
thing as an idea corresponding to "an elaboration of Gone With the
Wind," but once a reader has the embodiment of that idea in their
hands, a work subject to copyright has obviously been created, and
it's up to the courts to decide whether it's a derived work or not.

Linux is a copyrighted work, so "A Foo2000 SCSI driver for Linux 2.6,"
once it gets embodied in software, unless it's implemented in
userspace, is most likely going to be a work derived from the
copyrighted expression which is the linux kernel.

I don't think a software company is going to get away with declaring
that their driver is parody (though I've seen code that appears to be
a parody of computer programming generally), and if they're careful
enough not to use the same symbols as me... well, they won't have
created a derived work, but  other important senses of the word "work"
probably won't apply to their program, either.

Anyway, since I'm not a lawyer and no one is suing anyone, I'm really,
really done now.

Night,
Dave

RE: [future of drivers?] a proposal for binary drivers.

[David Schwartz]

> On 3/9/06, David Schwartz <davids@webmaster.com> wrote:

> > > Copyright law also absolutely
> > > gives me monopoly power over derivitive works - you can't even create
> > > a work derived from my dropped-from-airplane poem without my
> > > permission, much less distribute one.
> >
> >        No, that's not true on two counts.
> >
> >        First, your last part "much less distribute one" is
> > utterly false.
> > Copyright law does not give you any special rights to restrict the
> > distribution of derivative works.

> You're being rather hyper-technical and semantic here, aren't you? You
> clearly can't distribute something you can't legally "prepare" in the
> first place.

	No, it's a huge point. Because there are other ways you can get the right
to *prepare* it, these include fair use, first sale, scenes a faire,
necessary step, de minimis, and probably several other things I'm missing.

> > Second, your first part, that it gives you monopoly
> > power over the creation
> > of derivative works is also false. First sale and fair use can
> > give people
> > the right to create derivative works.

> OK, 14 USC 106(2) gives me monopoly power, subject to certain
> restrictions. You seemed to have missed my point, as none of the
> restrictions you mentioned include creation of derivative software
> programs for commercial distribution.

	It's not clear that necessary step and scenes a faire don't. There is case
law regarding necessary step creation of derivative works for commercial
distribution. So you're simply assuming as beyond controversy an issue that
is totally unresolved.

> In another reply, you wrote:

> >        You cannot copyright an idea. "A Foo2000 SCSI driver for
> > Linux 2.6" is an
> > *idea*. So you cannot argue that you have copyright on every practically
> > possible way to create such a driver.

> Your argument, if extended to fiction, is equivalent to "'An
> elaboration of Gone With the Wind' is an idea."

	An elaboration of Gone With the Wind is not functional. It doesn't *do*
anything. So whether or not it's the only way to do a particular thing is
not even an issue.

> There may be such a
> thing as an idea corresponding to "an elaboration of Gone With the
> Wind," but once a reader has the embodiment of that idea in their
> hands, a work subject to copyright has obviously been created, and
> it's up to the courts to decide whether it's a derived work or not.

	Of course, and if all it took from the original work is the only way to get
a particular thing done, it's clearly not a derivate work for copyright
purposes.

> Linux is a copyrighted work, so "A Foo2000 SCSI driver for Linux 2.6,"
> once it gets embodied in software, unless it's implemented in
> userspace, is most likely going to be a work derived from the
> copyrighted expression which is the linux kernel.

	I agree, however it is not legally a derivative work if all it took from
Linux was the only (practical) way to get a particular functional task done.
This is scenes a faire. (Did you read Lexmark v. Static Controls?)

> I don't think a software company is going to get away with declaring
> that their driver is parody (though I've seen code that appears to be
> a parody of computer programming generally), and if they're careful
> enough not to use the same symbols as me... well, they won't have
> created a derived work, but  other important senses of the word "work"
> probably won't apply to their program, either.

	The symbols are not the issue. Courts have clearly held that the names of
symbols and the order of functions is not copyrightable content. Anything
that is purely functional is not copyrightable content either. If it is the
only, straightforward, practical way to do a particular thing, it's just not
copyrightable. (If you want software patents, you know where to find them.)

	Software is not the same as other works, and the analogies to
non-functional works don't always, excuse the term, work.

	DS

Re: [future of drivers?] a proposal for binary drivers.

[Helge Hafting]

Anshuman Gholap wrote:

>well ya, I knew i was running the risk to be labelled like that, cause
>i thought to talk of this issue, more shake is needed that just stir.
>
>please dont get me wrong (even though i think most of you already
>have), i own my graditude for the livelihood i am having to
>linux,linus and co.
>  
>
>this discussion is totally for betterment of the new users who should
>not be forced to become developers in order to get a trival thing
>running on their desktop/laptop, like device driver.
>  
>
The new user who don't want to be developer should
get his linux from a linux distributor, who will
provide _all_ supported drivers auto-installed
during a normal install.  No development by the user,
no compiling. 

Helge Hafting

Re: [future of drivers?] a proposal for binary drivers.

[Luke-Jr]

On Wednesday 08 March 2006 09:35, Anshuman Gholap wrote:
> 1) The people who have control over linux kernel as of now are
> countable on hands but cannot be held countable for work cause they do
> it as hobby/"insert anything which says working for free", now for a
> peice of code like linux kernel, such kind of aloofness regarding
> manpower and kind_of nazism in not allowing others to dynamically
> get_work_done (like binary driver) seems totally wrong.

Binary drivers are no more "work" than robbing a bank is. It is making a 
profit through cheating people, which is of course what is truly wrong.

> 2) there are two possibilities here, a) linus and co can gather in a
> building pay all intellects and allow fast driver developments, b)
> allow binary drivers to work with linux kernel dynamically(with their
> own license what they choose).

Or Linux can remain GPL'd, which prohibits binary drivers *legally*, and back 
this by keeping a non-stable API which prohibits binary drivers 
*technically*.

Re: [future of drivers?] a proposal for binary drivers.

[Phillip Susi]

Luke-Jr wrote:
> Or Linux can remain GPL'd, which prohibits binary drivers *legally*, and back 
> this by keeping a non-stable API which prohibits binary drivers 
> *technically*.

If binary drivers are illegal, then why have ATI and nvidia not been 
sued yet?

Interacting with the kernel does not make your software a derived work. 
  A derived work is if you make your own kernel that is very close to a 
straight copy of the Linux kernel.  The right to create new works that 
interact with others ( and therefore, require some understanding of how 
the other work operates ) is specifically protected by the US copyright 
act.

This is why it is legal to reverse engineer a binary driver to gain an 
understanding of how the hardware operates, publish that information, 
and then use that information to create new software to operate that 
hardware.

Re: [future of drivers?] a proposal for binary drivers.

[Olivier Galibert]

On Thu, Mar 09, 2006 at 11:29:08AM -0500, Phillip Susi wrote:
> This is why it is legal to reverse engineer a binary driver to gain an 
> understanding of how the hardware operates, publish that information, 
> and then use that information to create new software to operate that 
> hardware.

Hardware is not a copyrightable work.

  OG.

Re: [future of drivers?] a proposal for binary drivers.

[Dave Neuer]

On 3/9/06, Phillip Susi <psusi@cfl.rr.com> wrote:
> Luke-Jr wrote:
> > Or Linux can remain GPL'd, which prohibits binary drivers *legally*, and back
> > this by keeping a non-stable API which prohibits binary drivers
> > *technically*.
>
> If binary drivers are illegal, then why have ATI and nvidia not been
> sued yet?

Because no sufficiently deep-pocketed plaintiff has chosen to do so
yet. Don't incorrectly infer anything about the existence of a cause
of action from a lack of legal proceedings so far.

>
> Interacting with the kernel does not make your software a derived work.

That may or may not be true, depending on the nature of the
interaction, and the arbiter of truth in this case is the court
system, not you or I.

>  A derived work is if you make your own kernel that is very close to a
> straight copy of the Linux kernel.  The right to create new works that
> interact with others ( and therefore, require some understanding of how
> the other work operates ) is specifically protected by the US copyright
> act.

There are no dearth of legal opinions on this matter which differ
quite radically from your interpretation here, quite a few from
lawyers. As far as I am concerned (and the GPL too, if my
interpretation of it is correct), any code is a derived work of my
code if either a) it directly makes use of symbols in my code or b) it
cannot execute unless my code executes, such that its distribution
without my code would be useless.

>
> This is why it is legal to reverse engineer a binary driver to gain an
> understanding of how the hardware operates, publish that information,
> and then use that information to create new software to operate that
> hardware.

No, you are referring to a restriction on the limitations in software
licenses which is separate from copyright. Copyright law does not talk
about interoperability at all. And even the applicability of the
restriction to which you refer is jurisdiction-dependant as well as
context-dependant (see the DMCA).

Regards,
Dave

Re: [future of drivers?] a proposal for binary drivers.

[Xavier Bestel]

Le jeudi 09 mars 2006 à 12:33 -0500, Dave Neuer a écrit :
> On 3/9/06, Phillip Susi <psusi@cfl.rr.com> wrote:
> > If binary drivers are illegal, then why have ATI and nvidia not been
> > sued yet?
> 
> Because no sufficiently deep-pocketed plaintiff has chosen to do so
> yet.

No. It's just because they don't distribute a kernel with their drivers.

	Xav

Re: [future of drivers?] a proposal for binary drivers.

[Dave Neuer]

On 3/9/06, Xavier Bestel <xavier.bestel@free.fr> wrote:
> Le jeudi 09 mars 2006 à 12:33 -0500, Dave Neuer a écrit :
> > On 3/9/06, Phillip Susi <psusi@cfl.rr.com> wrote:
> > > If binary drivers are illegal, then why have ATI and nvidia not been
> > > sued yet?
> >
> > Because no sufficiently deep-pocketed plaintiff has chosen to do so
> > yet.
>
> No. It's just because they don't distribute a kernel with their drivers.

IF their driver is a derivative work (which many have argued), then it
does not matter that they don't distribute the kernel (which they
would have a perfect right to do, like you or I or anyone else, under
the terms of the GPL); what matters is that copyright law prevents the
CREATION of derivative works without permission, and the GPL states
that permission in this case is contingent upon distributing (*) said
derivative works under the GPL (i.e., full source code availability).

Thus, if binary modules are in fact derivative works, ATI and NVidia
are not legally allowed to distribute them. You may disagree about
whether or not drivers are derivative works; to my knowledge no court
has ruled on this yet.

But I stand by my assertion: many kernel developers on record stating
that they don't want their work used in binary-only modules, and the
reason that this hasn't been decided by a court yet is no sufficiently
deep-pocketed plaintiff (independantly wealthy kernel hackers or a big
corporation with copyright interest in the kernel) has decided to sue,
yet.

See Linus' statements here:
http://www.ussg.iu.edu/hypermail/linux/kernel/0312.0/0670.html
and here: http://www.ussg.iu.edu/hypermail/linux/kernel/0312.1/0708.html
if you think I'm just pulling this stuff out of my butt.

Regards,
Dave

* the GPL doesn't say you can't use modified or derived GPL software
internally, so ATI and nVidia would presumably be OK if they didn't
distribute the drivers.

Re: [future of drivers?] a proposal for binary drivers.

[marty fouts]

> But I stand by my assertion: many kernel developers on record stating
> that they don't want their work used in binary-only modules, and the
> reason that this hasn't been decided by a court yet is no sufficiently
> deep-pocketed plaintiff (independantly wealthy kernel hackers or a big
> corporation with copyright interest in the kernel) has decided to sue,
> yet.
>
> See Linus' statements here:
> http://www.ussg.iu.edu/hypermail/linux/kernel/0312.0/0670.html
> and here: http://www.ussg.iu.edu/hypermail/linux/kernel/0312.1/0708.html
> if you think I'm just pulling this stuff out of my butt.
>

Looking at Linus' arguments, how would you say those kernel developers
feel about the following scenario:

I have access to a 3rd party file system, written not for Linux but
for some completely different OS. But my license with that vendor does
not allow me to distribute the file system.   I write the translation
layer that they describe in their documentation that allows me to drop
their file system, unchanged, into Linux. I GPL the translation layer
and make the source available appropriately. (This is similar to the
AFS point in Linus' email, but not exactly the same.) I do not, since
I don't permission to, distribute the source for the third party OS.

1) Have I met my legal obligation under the GPL? (Seems to me Linus
would say yes, but I want to understand other people's view on this.)

2) Will the developers you mention above be unhappy anyway, even if I have?

Re: [future of drivers?] a proposal for binary drivers.

[Eduard Bloch]

#include <hallo.h>
* Dave Neuer [Thu, Mar 09 2006, 03:22:29PM]:

> But I stand by my assertion: many kernel developers on record stating
> that they don't want their work used in binary-only modules, and the
> reason that this hasn't been decided by a court yet is no sufficiently
> deep-pocketed plaintiff (independantly wealthy kernel hackers or a big
> corporation with copyright interest in the kernel) has decided to sue,
> yet.

Strongly following this attitude they need to expand the current
MODULE_LICENSE insanity into userspace. The are lots of evil powers,
using non-GPL-ed software which calls the holy GPLed syscall handlers.

Honestly, using double standards this way sucks. The last strike I
stubled over was putting class device registration methods into the GPL
"enforcing" macros, cloacked as mandatory interface changes.

This is no longer funny, it is on the same level as drawing a line
around the toilet box and saying: you cannot use it any longer, go and
piss somewhere else even if you have to search two hours for a suitable
place.

Eduard.

-- 
Captain John Sheridan: No surrender, no retreat.
                                                 -- Quotes from Babylon 5 --

Re: [future of drivers?] a proposal for binary drivers.

[Jan Knutar]

On Saturday 11 March 2006 02:54, Eduard Bloch wrote:

> This is no longer funny, it is on the same level as drawing a line
> around the toilet box and saying: you cannot use it any longer, go and
> piss somewhere else even if you have to search two hours for a suitable
> place.

That is not entirely unusual situation in real life ;-)

Re: [future of drivers?] a proposal for binary drivers.

[DervishD]

    Hi Eduard :)

 * Eduard Bloch <edi@gmx.de> dixit:
> This is no longer funny, it is on the same level as drawing a line
> around the toilet box and saying: you cannot use it any longer, go
> and piss somewhere else even if you have to search two hours for a
> suitable place.

    No, it's not on the same level. It's on the same level as giving
your thesis draft to a comrade so he can study and pass his exams,
and in turn he publish the thesis as his and charge money for that.

    I don't want my work used by a corporation without giving any
modification under the same conditions under I published my work.
Binary driver can and will do harm if allowed.

    Raúl Núñez de Arenas Coronado

-- 
Linux Registered User 88736 | http://www.dervishd.net
http://www.pleyades.net & http://www.gotesdelluna.net
It's my PC and I'll cry if I want to... RAmen!

Re: [future of drivers?] a proposal for binary drivers.

[Eduard Bloch]

#include <hallo.h>
* DervishD [Sat, Mar 11 2006, 10:16:23AM]:
>     Hi Eduard :)
> 
>  * Eduard Bloch <edi@gmx.de> dixit:
> > This is no longer funny, it is on the same level as drawing a line
> > around the toilet box and saying: you cannot use it any longer, go
> > and piss somewhere else even if you have to search two hours for a
> > suitable place.
> 
>     No, it's not on the same level. It's on the same level as giving
> your thesis draft to a comrade so he can study and pass his exams,
> and in turn he publish the thesis as his and charge money for that.
> 
>     I don't want my work used by a corporation without giving any
> modification under the same conditions under I published my work.

Ehm, I think you are overlook a serious issue assuming that every touch
of non-GPLed software immediately implies use and abuse of GPLed
software. That is not what I am talking about! I talk about keeping
compatible interfaces.

But where do you draw the line between communication and abuse of
other's work? What is the point of deliberate merge of interfaces with
code if not throwing stones in ways of others? Going back to your
comparison, it is like distributing your thesis with a following
license:

Noone ever making reference to my work is allowed to use the following
words: <followed by a list of domain specific terms>. Invent your own
words, but those are mine!!!!!1

Sounds childish? Exactly like GPLing interfaces is.

> Binary driver can and will do harm if allowed.

Welcome at the dark side. "Only the Sith deal in absolutes".

Eduard.

-- 
Captain John Sheridan: Delenn, I have been working up a good mad all day and I
am NOT about to let you ruin it by agreeing with me!
                                                 -- Quotes from Babylon 5 --

RE: [future of drivers?] a proposal for binary drivers.

[David Schwartz]

>     No, it's not on the same level. It's on the same level as giving
> your thesis draft to a comrade so he can study and pass his exams,
> and in turn he publish the thesis as his and charge money for that.

	Except it's not. We're not talking about people taking the Linux kernel and
then publishing and charging for the Linux kernel. If we were, your analogy
would be correct.

	What you're talking about is more like you giving your thesis draft to a
comrade so that he can study and pass his exams and then claiming that you
own his exams. Or that you own his criticisms to your draft.

>     I don't want my work used by a corporation without giving any
> modification under the same conditions under I published my work.
> Binary driver can and will do harm if allowed.

	If you want to restrict *use* you need an EULA, shrink wrap agreement,
click-through or signed contract. If you give away copies of your work with
no conditions on the *receipt* of the work, you lose the right to control
how the work is used. Otherwise, someone could drop a million copies of
their poem from an airplane and then sue everyone who read it.

	Copyright is simply not powerful enough to allow you to control *any*
practical way to do a particular thing (say, make an NE2000 card work with
Linux 2.6). It is only powerful enough to allow you to control the one
specific way that *you* chose to do something. If you want software patents,
you know where to find them.

	You cannot use copyright to own *every* way to express a particular idea.
You cannot even use it to own every practical way to express a particular
idea. Is is quite clear that "use a different operating system" or "use
hardware for which there are already drivers in the kernel" are simply other
ideas, not other ways to express the same idea.

	I advise everyone with an interest to read carefully the entire decision in
Lexmark v. Static Controls. It clearly talks about how once you own every
practical way to do a particular thing, you cease to be allowed to use
copyright to do it. (Lexmark had a copyrighted Toner Loading Program in
their print cartridges which Static Controls 'stole' to make compatible
print cartridges. The court held that, among other things, even though the
TLP would otherwise have been copyrightable, since it was the only practical
way to make a cartridge work with certain Lexmark printers, copyright was
not applicable.)

	DS

Re: [future of drivers?] a proposal for binary drivers.

[DervishD]

    Hi David :)

 * David Schwartz <davids@webmaster.com> dixit:
> >     I don't want my work used by a corporation without giving any
> > modification under the same conditions under I published my work.
> > Binary driver can and will do harm if allowed.
> 
> 	If you want to restrict *use* you need an EULA, shrink wrap agreement,
> click-through or signed contract. If you give away copies of your work with
> no conditions on the *receipt* of the work, you lose the right to control
> how the work is used. Otherwise, someone could drop a million copies of
> their poem from an airplane and then sue everyone who read it.

    Sorry, I meant "make a derivative work and distribute it" when I
wrote "used by a corporation without giving any modification...".
My english is very poor sometimes O:)

    I was referring to the fact that if I use GPL is because I don't
want anyone using my work to produce new work and distribute it
without distributing the modification, too. In the kernel case, a
binary driver uses work made by others without giving anything back
(and not, I don't consider the driver itself enough "giving back"),
at least that's how I see it.

    If binary drivers are allowed, soon we will have only drivers for
a couple of distros (I don't use a distro, so I'm lost) and they will
be unmaintained as soon as new hardware is released. I have had that
problem in Windows with hardware that is only three years old,
hardware that I can use in Linux without problem (my Linux box is 5
years old on the average, but my graphics card was manufactured in
1998 IIRC). In MS-DOS, binary drivers were an issue because they were
abandoned as soon as Windows-95 was released, but the worst thing is
that a good bunch of GOOD hardware will work ONLY with the latest
release of WinXP. I don't want that to happen in Linux. And if I have
the sources, I have a chance of fixing bugs or whatever.

    I know copyright won't help in that issue, but licensing can, and
I think that the kernel is doing the right thing.

    Raúl Núñez de Arenas Coronado

-- 
Linux Registered User 88736 | http://www.dervishd.net
http://www.pleyades.net & http://www.gotesdelluna.net
It's my PC and I'll cry if I want to... RAmen!

Re: [future of drivers?] a proposal for binary drivers.

[Dave Neuer]

On 3/11/06, David Schwartz <davids@webmaster.com> wrote:
>
>         I advise everyone with an interest to read carefully the entire decision in
> Lexmark v. Static Controls. It clearly talks about how once you own every
> practical way to do a particular thing, you cease to be allowed to use
> copyright to do it. (Lexmark had a copyrighted Toner Loading Program in
> their print cartridges which Static Controls 'stole' to make compatible
> print cartridges. The court held that, among other things, even though the
> TLP would otherwise have been copyrightable, since it was the only practical
> way to make a cartridge work with certain Lexmark printers, copyright was
> not applicable.)

Static Controls also explicitly says that the analysis of whether
scenes a faire applies is vastly different for a work of greater
complexity and size than the TLP ("Neither do the cited cases support
the district court's initial frame of reference. [cases cited],
involved copies of Apple's operating system program -- a program whose
size and complexity is to the Toner Loading Program what the Sears
Tower is to a lamppost. Given the nature of the Apple program, it
would have been exceedingly difficult to say that practical
alternative means of expression did not exist...").

Thanks for the citation and the legal analysis, David. I've found your
posts insightful and educational, but after reading more I still come
away thinking you've missed the essential point that the Linux
operating system, and in particular the means that internal parts of
the system communicate with each other (essential to modules) are
inherently part of the expressive content of Linux, not dictated by
external factors, and that modules are indeed derived works of the
kernel and hence subject to infringement claims if the license terms
aren't met.

Ciao,
Dave

Re: [future of drivers?] a proposal for binary drivers.

[David Schwartz]

Sorry for any format issues, I'm not able to use my normal email client 
right now.

> Static Controls also explicitly says that the analysis of whether
> scenes a faire applies is vastly different for a work of greater
> complexity and size than the TLP ("Neither do the cited cases support
> the district court's initial frame of reference. [cases cited],
> involved copies of Apple's operating system program -- a program whose
> size and complexity is to the Toner Loading Program what the Sears
> Tower is to a lamppost. Given the nature of the Apple program, it
> would have been exceedingly difficult to say that practical
> alternative means of expression did not exist...").

The whole issue here is people who claim that copyright allows them to own 
*any* way to make an NE2000 network card work with linux v2.6. How can you 
say that practical alternative means exist if the claim is that every such 
way is owned?

You can use hardware that's already supported in the kernel. You can use 
other operating systems than Linux. But Static Controls could also have made 
cartridges for other printers or printers from other manufacturers.

The issue is not the complexity of the TLP, the issue is simply that you 
cannot use copyright to get protection that is capable of being expressed in 
functional terms. You cannot own every way to express a functional idea. 
That's what patents are for.

DS

Re: [future of drivers?] a proposal for binary drivers.

[Dave Neuer]

On 3/12/06, David Schwartz <davids@webmaster.com> wrote:
>
> The issue is not the complexity of the TLP, the issue is simply that you
> cannot use copyright to get protection that is capable of being expressed in
> functional terms. You cannot own every way to express a functional idea.
> That's what patents are for.

That is not what the court said, and the section I quoted you shows
that quite plainly.

In Static Controls, the issue was a 55 byte program to calculate the
level of toner in a cartridge. The court ruled that the program design
of the TLP was so constrained by external factors (the efficient
execution of a small number of calculations) that any other
implementation would have been impractical.

Linux is a completely different matter, directly analogous to Apple's
OS in the court's analysis. There are no such external factors
dictating the form of the kernel's facilities for integrating new
functionality. The kernel developers could have chosen some other
means for drivers to coordinate their activities with the kernel than
the current driver model (for instance, the means employed in Linux
2.4).

You keep insisting that "a driver for hardware X under Linux" is a
functional idea. It is not. "Calculate the amount of toner left" is a
functional idea. "Set the control register of hardware X to value Y"
is a functional idea (and not copyrightable due to scenes a faire).

I can understand how you might think, "well, nVidia could have chosen
some other means of representing pixels on the screen and controlling
them" is analagous to "well, the kernel developers could have chosen
some other means for modules or drivers to interact with the system,"
but graphics hardware is not copyrightable, software is. To conflate
the two as you seem to want to do would render pretty much all
software uncopyrightable. That might be preferable to you, but it
would crush innovation in software development and make it impossible
for anyone but largish businesses to create software.

Dave

RE: [future of drivers?] a proposal for binary drivers.

[David Schwartz]

> In Static Controls, the issue was a 55 byte program to calculate the
> level of toner in a cartridge. The court ruled that the program design
> of the TLP was so constrained by external factors (the efficient
> execution of a small number of calculations) that any other
> implementation would have been impractical.

	Exactly. And this is precisely what is happening here. The kernel headers
are small in comparison to the kernel. And external factors are such that
there is no other way to create kernel modules other than by using the
kernel headers.

> Linux is a completely different matter, directly analogous to Apple's
> OS in the court's analysis. There are no such external factors
> dictating the form of the kernel's facilities for integrating new
> functionality.

	You are saying there are practical ways to develop kernel modules other
than using the kernel headers?

> The kernel developers could have chosen some other
> means for drivers to coordinate their activities with the kernel than
> the current driver model (for instance, the means employed in Linux
> 2.4).

	Sure, and Lexmark could have allowed the TLP to be a million bytes.

> You keep insisting that "a driver for hardware X under Linux" is a
> functional idea. It is not. "Calculate the amount of toner left" is a
> functional idea. "Set the control register of hardware X to value Y"
> is a functional idea (and not copyrightable due to scenes a faire).

	If that was true, then why didn't the court say to Static Controls
"calculate the toner left on some other printer, you cannot take the TLP"?

> To conflate
> the two as you seem to want to do would render pretty much all
> software uncopyrightable. That might be preferable to you, but it
> would crush innovation in software development and make it impossible
> for anyone but largish businesses to create software.

	Not at all. The only case where software is uncopyrightable is when it
claims to cover *every* practical way to make X do Y with Z. Just as the TLP
was the only (practical) way to measure the toner left in particular Lexmark
printers.

	DS

Re: [future of drivers?] a proposal for binary drivers.

[Helge Hafting]

David Schwartz wrote:

>  
>
>>In Static Controls, the issue was a 55 byte program to calculate the
>>level of toner in a cartridge. The court ruled that the program design
>>of the TLP was so constrained by external factors (the efficient
>>execution of a small number of calculations) that any other
>>implementation would have been impractical.
>>    
>>
>
>	Exactly. And this is precisely what is happening here. The kernel headers
>are small in comparison to the kernel. And external factors are such that
>there is no other way to create kernel modules other than by using the
>kernel headers.
>  
>
Smaller than the kernel, but still big.  These are not 55 bytes.
And the kernel headers are not unique either.  You can make
many changes and still be able to use them for a driver.

>>Linux is a completely different matter, directly analogous to Apple's
>>OS in the court's analysis. There are no such external factors
>>dictating the form of the kernel's facilities for integrating new
>>functionality.
>>    
>>
>
>	You are saying there are practical ways to develop kernel modules other
>than using the kernel headers?
>  
>
Many of these interfaces are published.  Buy a linux device driver
book and write your own headers - possibly with different names
for the types involved.  Structs or just a bunch of variables in a
certain order?  Your call again.  You won't get everything this way,
but you don't need everything to write one driver.

Then there is reverse engineering.  More work, but certainly
feasible seeing how many linux drivers were written this way.

Helge Hafting

Re: [future of drivers?] a proposal for binary drivers.

[Dave Neuer]

On 3/10/06, Eduard Bloch <edi@gmx.de> wrote:
>
> Strongly following this attitude they need to expand the current
> MODULE_LICENSE insanity into userspace. The are lots of evil powers,
> using non-GPL-ed software which calls the holy GPLed syscall handlers.
>
> Honestly, using double standards this way sucks. The last strike I
> stubled over was putting class device registration methods into the GPL
> "enforcing" macros, cloacked as mandatory interface changes.
>
> This is no longer funny, it is on the same level as drawing a line
> around the toilet box and saying: you cannot use it any longer, go and
> piss somewhere else even if you have to search two hours for a suitable
> place.

I'm not sure how the analogy of a toilet fits either the linux kernel
or the idea that I'm somehow obligated to work without compensation so
some corporation can make a profit I get no share in.

Dave

RE: [future of drivers?] a proposal for binary drivers.

[David Schwartz]

> I'm not sure how the analogy of a toilet fits either the linux kernel
> or the idea that I'm somehow obligated to work without compensation so
> some corporation can make a profit I get no share in.

	You are not obligated to work without compensation or to let some
corporation use your work to make a profit that you get no share in.
However, if you choose to give your work away without first getting an
agreement from the recipient, you are willingly giving up lots of control
that you would otherwise have.

	I would strongly caution you against believing anyone who tells you
different, no matter how much you want to hear it. The facts are:

	1) A person who lawfully acquires a work without agreeing otherwise gains
the right to the ordinary and expected use of that work.

	2) The ordinary and expected use of a library is to produce applications
that use that library.

	3) The ordinary and expected use of the RedHat 'linux-kernel' package is to
develop kernel drivers and produce binaries of them.

	4) Copyright does not allow you to own every way to do some specific thing,
you need a patent for that. Any application that uses library X or any
driver for kernel Y is a specific thing. Copyright only applies when there
are numerous ways to do the same thing or express the same idea. Drivers for
different operating systems are different ideas. You cannot use copyright to
lock out someone from doing a particular thing, only from doing that thing
the same way you did.

	5) There is no right under copyright for authors of original works to limit
the distribution of lawfully-created derivative works to those with the
right to use the original work.

	6) All of this is copyright law and applies whether or not anyone agrees to
the GPL or any other agreement, so nothing those agreements says can change
this.

	DS

RE: [future of drivers?] a proposal for binary drivers.

[David Schwartz]

> 	3) The ordinary and expected use of the RedHat 
> 'linux-kernel' package is to develop kernel drivers and produce 
> binaries of them.

	Sorry, that should of course read 'kernel-devel'.

	DS

RE: [future of drivers?] a proposal for binary drivers.

[Arjan van de Ven]

> 	2) The ordinary and expected use of a library is to produce applications
> that use that library.

and the kernel is not a library

> 	5) There is no right under copyright for authors of original works to limit
> the distribution of lawfully-created derivative works to those with the
> right to use the original work.

this is not correct. Copyright law *DOES* give copyright owners the
right to control derived works. 

RE: [future of drivers?] a proposal for binary drivers.

[David Schwartz]

> > 	2) The ordinary and expected use of a library is to produce
> > applications
> > that use that library.

> and the kernel is not a library

	Tell me, then, what is the orindary expected use of the RedHat
'kernel-devel' package?

> > 	5) There is no right under copyright for authors of
> > original works to limit
> > the distribution of lawfully-created derivative works to those with the
> > right to use the original work.

> this is not correct. Copyright law *DOES* give copyright owners the
> right to control derived works.

	No, it does not. Copyright law only gives copyright owners the right to
control the *creation* of derivative works. I very carefully worded my
statement above so that it would talk about precisely the right people claim
they have and precisely the right they do not have.

	In this case, the alleged derivative work is created under first sale, as
part of normal use. It is impossible to normally use the 'kernel-devel'
package without creating derivative works, and under first sale, normal use
(and anything required for normal use) cannot be burdened. Once the
derivative work is lawfully created, there is nothing in copyright law that
requires the permission of the author of the original work to distribute the
derived work to licensees of the original work.

	The GPL gets around this by imposing requirements on the creation of
derivative works, under the assumption that you cannot get the right to
create a derivative work any other way. But this is false, first sale grants
the right to normal use, and normal use includes anything necessary for
normal use. For a library or for the 'kernel-devel' package, normal use
requires the creation of derivative works.

	DS

Re: [future of drivers?] a proposal for binary drivers.

[sean]

On Mon, 13 Mar 2006 13:57:40 -0800
"David Schwartz" <davids@webmaster.com> wrote:

> 	No, it does not. Copyright law only gives copyright owners the right to
> control the *creation* of derivative works. I very carefully worded my
> statement above so that it would talk about precisely the right people claim
> they have and precisely the right they do not have.
> 
> 	In this case, the alleged derivative work is created under first sale, as
> part of normal use. It is impossible to normally use the 'kernel-devel'
> package without creating derivative works, and under first sale, normal use
> (and anything required for normal use) cannot be burdened. Once the
> derivative work is lawfully created, there is nothing in copyright law that
> requires the permission of the author of the original work to distribute the
> derived work to licensees of the original work.
> 
> 	The GPL gets around this by imposing requirements on the creation of
> derivative works, under the assumption that you cannot get the right to
> create a derivative work any other way. But this is false, first sale grants
> the right to normal use, and normal use includes anything necessary for
> normal use. For a library or for the 'kernel-devel' package, normal use
> requires the creation of derivative works.
> 

So i buy a book; clearly the reason it's written in english is so that I can
extend and alter it.   So I rip out the last chapter and replace it with one
of my own.   Now _clearly_ i can distribute this new work around the world
without any fear of being sued by the copyright holders because it's fair use.
NOT!

Now before you try to argue that altering copyrighted source code is fair
use but altering copyrighted books isn't; just stop.  Please leave this 
matter to the lawyers and off this list.

Sean

RE: [future of drivers?] a proposal for binary drivers.

[David Schwartz]

> On Mon, 13 Mar 2006 13:57:40 -0800
> "David Schwartz" <davids@webmaster.com> wrote:

> > 	No, it does not. Copyright law only gives copyright owners
> > the right to
> > control the *creation* of derivative works. I very carefully worded my
> > statement above so that it would talk about precisely the right
> > people claim
> > they have and precisely the right they do not have.

> > 	In this case, the alleged derivative work is created under
> > first sale, as
> > part of normal use. It is impossible to normally use the 'kernel-devel'
> > package without creating derivative works, and under first
> > sale, normal use
> > (and anything required for normal use) cannot be burdened. Once the
> > derivative work is lawfully created, there is nothing in
> > copyright law that
> > requires the permission of the author of the original work to
> > distribute the
> > derived work to licensees of the original work.

> > 	The GPL gets around this by imposing requirements on the creation of
> > derivative works, under the assumption that you cannot get the right to
> > create a derivative work any other way. But this is false,
> > first sale grants
> > the right to normal use, and normal use includes anything necessary for
> > normal use. For a library or for the 'kernel-devel' package, normal use
> > requires the creation of derivative works.
> >

> So i buy a book; clearly the reason it's written in english is so
> that I can
> extend and alter it.

	How is that clear? The normal use of a book is to read it and think about
it. Extending and altering it is not only not the normal use but explicitly
protected under copyright law (as a sequel).

	While making a kernel module can be argued to be extending the kernel, I
don't think you can argue it's "altering" the kernel. I certainly would not
argue that you can produce an altered version of the kernel and distribute
it in binary only form.

> So I rip out the last chapter and replace
> it with one
> of my own.   Now _clearly_ i can distribute this new work around the world
> without any fear of being sued by the copyright holders because
> it's fair use.
> NOT!

	Again, your example fails because you are talking about distributing it to
people who do not lawfully acquire the original work. You notice I carefully
said "to licensees of the original work" and yet your example talks about
distribute to people who are not licensees of the original work.

	A better example might be if you and a friend each buy a copy of music. You
transpose it into a different key for your friend and then give him a copy
of the transposed work. There's a huge difference between giving that copy
to a licensee of the original work and giving it to someone who is not a
licensee of the original work. You can argue whether you could lawfully
transpose it in the first place, but assuming the creation is lawful, you
can certainly distribute it to other licensees of the original work. (No
court, AFAIK, has ever held otherwise and nothing in copyright law says
otherwise. Please correct me if I'm wrong.)

> Now before you try to argue that altering copyrighted source code is fair
> use but altering copyrighted books isn't; just stop.

	I wasn't talking about fair use. If you think I would argue fair use, you
really have no idea what I'm talking about. I'm talking about:

	1) First sale. That is, the right to the normal use of a lawfully-acquired
work, and

	2) The absence of a specific right under copyright to restrict the
distribution of lawfully-created derivative works.

	And who was ever talking about altering anything? Why would you think this
is about "altering" copyrighted source code? Making an application that uses
a library isn't about "altering" the library. Making a kernel module isn't
about "altering" the kernel. Perhaps you are replying on auto-pilot and not
reading what I'm actually writing. I am *NOT* talking about modifying a
GPL'd work and then distributing the modified work in binary form.

> Please leave this
> matter to the lawyers and off this list.

	If you want it off the list, don't reply or reply off the list. But if you
choose to reply with an argument, I have as much right to rebut it in a
given forum as you do to make it. How can a rebuttal of an argument be any
less on-topic than the argument?

	DS

Re: [future of drivers?] a proposal for binary drivers.

[Valdis.Kletnieks]

[-- Attachment #1: Type: text/plain, Size: 782 bytes --]

On Mon, 13 Mar 2006 09:20:41 +0100, Arjan van de Ven said:
> > 	5) There is no right under copyright for authors of original works to limit
> > the distribution of lawfully-created derivative works to those with the
> > right to use the original work.
> 
> this is not correct. Copyright law *DOES* give copyright owners the
> right to control derived works. 

He quite clearly said "*lawfully-created* derivative works".

http://www.law.cornell.edu/uscode/html/uscode17/usc_sec_17_00000106----000-.html

quite clearly says that the right to *produce* a derivative work can be controlled
by the copyright holder (17 USC 106(2)) but nothing in 17 USC 106 says that
the copyright holder can control the distribution of an *already-authorized*
derivative work.

IANAL and all that....

[-- Attachment #2: Type: application/pgp-signature, Size: 228 bytes --]

RE: [future of drivers?] a proposal for binary drivers.

[Bernd Petrovitsch]

On Sun, 2006-03-12 at 21:16 -0800, David Schwartz wrote:
> > I'm not sure how the analogy of a toilet fits either the linux kernel
> > or the idea that I'm somehow obligated to work without compensation so
> > some corporation can make a profit I get no share in.
> 
> 	You are not obligated to work without compensation or to let some
> corporation use your work to make a profit that you get no share in.
> However, if you choose to give your work away without first getting an
> agreement from the recipient, you are willingly giving up lots of control
> that you would otherwise have.
> 
> 	I would strongly caution you against believing anyone who tells you
> different, no matter how much you want to hear it. The facts are:
> 
> 	1) A person who lawfully acquires a work without agreeing otherwise gains
> the right to the ordinary and expected use of that work.

... under the given conditions/rules/license/contract/etc. If you
lawfully buy MSFT software, you are probably also limited by your
contract (as long as the given limitations do not conflict with laws
under the jurisdiction you bought and/or live - more or less IANAL).

[ irrelevant library stuff removed ]

> 	3) The ordinary and expected use of the RedHat 'linux-kernel' package is to
> develop kernel drivers and produce binaries of them.

Since you got the package from RedHat (or subsidiaries or several people
in between), you are bound by the GPL since neither RedHat nor the
several people in between as any different access to the kernel-sources
than via GPL. And they can't change it.
So you got it under the rules of the GPL.

> 	4) Copyright does not allow you to own every way to do some specific thing,

Copyright/authors rights allows me to own *my way* of doing it. And if
you derive your work on mine, it depends on the quality and quantity
*if* you have any copyright/authors rights on you patch and how the
joined work must be treated legally.
In the Linux kernel, dozens (if not hundreds) of people have
copyrighted/authored righted code in there so.

> you need a patent for that. Any application that uses library X or any
> driver for kernel Y is a specific thing. Copyright only applies when there

If there was a binary in-kernel API, yes.
But a) there is no "officially" one and b) we have no "libraries" (in
the sense of the GPL) here.

> are numerous ways to do the same thing or express the same idea. Drivers for
> different operating systems are different ideas. You cannot use copyright to
> lock out someone from doing a particular thing, only from doing that thing
> the same way you did.

No, you can't even lock someone out to do the same thing. You can only
lock someone out to base his thing on your thing (but you can't hinder a
reimplementation - you need a patent for this [and a jurisdiction which
allows software patents]).

> 	5) There is no right under copyright for authors of original works to limit
> the distribution of lawfully-created derivative works to those with the
> right to use the original work.

Maybe not under (US-American) copyright, but under continental European
authors rights there are such possibilities (I leave it to lawyers if
and when they apply) and you can't even remove them with contracts (as
with all law stuff).

> 	6) All of this is copyright law and applies whether or not anyone agrees to
> the GPL or any other agreement, so nothing those agreements says can change
> this.

This is a common misunderstanding: If you change the rules of the GPL,
you automatically loose all rights you received with the GPL[0]

	Bernd

[0]: If you believe the wording of the GPL. Lawyers may think different.
-- 
Firmix Software GmbH                   http://www.firmix.at/
mobil: +43 664 4416156                 fax: +43 1 7890849-55
          Embedded Linux Development and Services

RE: [future of drivers?] a proposal for binary drivers.

[Lee Revell]

On Mon, 2006-03-13 at 10:24 +0100, Bernd Petrovitsch wrote:
> Maybe not under (US-American) copyright, but under continental
> European authors rights there are such possibilities (I leave it to
> lawyers if and when they apply) and you can't even remove them with
> contracts (as with all law stuff).
> 

Please refrain from knee-jerk USA-bashing, you said yourself you don't
know the law.

Lee

RE: [future of drivers?] a proposal for binary drivers.

[Bernd Petrovitsch]

On Mon, 2006-03-13 at 12:16 -0500, Lee Revell wrote:
> On Mon, 2006-03-13 at 10:24 +0100, Bernd Petrovitsch wrote:
> > Maybe not under (US-American) copyright, but under continental
> > European authors rights there are such possibilities (I leave it to
> > lawyers if and when they apply) and you can't even remove them with
> > contracts (as with all law stuff).
> 
> Please refrain from knee-jerk USA-bashing, you said yourself you don't

Whatever "knee-jerk" means, I'm not "USA-bashing" as such - at most I'm
"copyright"-bashing (opposed to modern authors rights. Copyright is also
somewhat the predecessor of modern authors rights so there is room to
improve there.) and/or just reminding several people that US law is not
the only relevant in the world.

> know the law.

Yes, I'm not a lawyer, but I do know some parts of it enough to have an
opinion. And when it comes to details and special circumstances, the
specialists should speak up.

	Bernd
-- 
Firmix Software GmbH                   http://www.firmix.at/
mobil: +43 664 4416156                 fax: +43 1 7890849-55
          Embedded Linux Development and Services

RE: [future of drivers?] a proposal for binary drivers.

[David Schwartz]

> Since you got the package from RedHat (or subsidiaries or several people
> in between), you are bound by the GPL since neither RedHat nor the
> several people in between as any different access to the kernel-sources
> than via GPL. And they can't change it.
> So you got it under the rules of the GPL.

	This is a common misconception. The GPL *never* bings the recipient of a
GPL'd work until and unless he agrees to it. You can find this made quite
clear on the FSF's web site.

> > 	4) Copyright does not allow you to own every way to do some
> specific thing,

> Copyright/authors rights allows me to own *my way* of doing it. And if
> you derive your work on mine, it depends on the quality and quantity
> *if* you have any copyright/authors rights on you patch and how the
> joined work must be treated legally.
> In the Linux kernel, dozens (if not hundreds) of people have
> copyrighted/authored righted code in there so.

	The claim here is not that one way to make an NE2000 work with Linux 2.6 is
owned but that *every* way to do that is owned. This is impossible under
copyright. You can certainly own one specific way to make an NE2000 work
with linux 2.6 (that would be a copyrighted driver), but you cannot use
copyright to prevent anyone from implementing a particular function.

> > you need a patent for that. Any application that uses library X or any
> > driver for kernel Y is a specific thing. Copyright only applies
> > when there

> If there was a binary in-kernel API, yes.
> But a) there is no "officially" one and b) we have no "libraries" (in
> the sense of the GPL) here.

	It doesn't matter. You cannot protect a function with copyright. If there
is only one practical way to get a particular function, nobody can use
copyright to own it. Read Lexmark v. Static Controls or goole for "copyright
lock out".

> > are numerous ways to do the same thing or express the same
> > idea. Drivers for
> > different operating systems are different ideas. You cannot use
> > copyright to
> > lock out someone from doing a particular thing, only from doing
> > that thing
> > the same way you did.

> No, you can't even lock someone out to do the same thing. You can only
> lock someone out to base his thing on your thing (but you can't hinder a
> reimplementation - you need a patent for this [and a jurisdiction which
> allows software patents]).

	You cannot lock someone out from basing their think on your thing, if that
is the only practical way to express a particular idea. See Lexmark v.
Static Controls, among other cases.

> > 	6) All of this is copyright law and applies whether or not
> > anyone agrees to
> > the GPL or any other agreement, so nothing those agreements
> > says can change
> > this.

> This is a common misunderstanding: If you change the rules of the GPL,
> you automatically loose all rights you received with the GPL[0]

	Huh? I'm not changing anything. And I'm not talking about any rights
received with the GPL, I'm talking about rights granted by law under first
sale and scenes a faire.

	DS

Re: [future of drivers?] a proposal for binary drivers.

[Phillip Susi]

Dave Neuer wrote:
>> Interacting with the kernel does not make your software a derived work.
> 
> That may or may not be true, depending on the nature of the
> interaction, and the arbiter of truth in this case is the court
> system, not you or I.
> 

That is correct, the final decision is up to a Judge, but we aren't in 
court here, we're discussing it, and so I'm making an argument that I 
believe any sane judge would side with.

> There are no dearth of legal opinions on this matter which differ
> quite radically from your interpretation here, quite a few from
> lawyers. As far as I am concerned (and the GPL too, if my
> interpretation of it is correct), any code is a derived work of my
> code if either a) it directly makes use of symbols in my code or b) it
> cannot execute unless my code executes, such that its distribution
> without my code would be useless.
> 

As you said above, your opinion doesn't matter, only the law and a 
judge's interpretation of it, and the law says no such thing.  Take a 
look at USC Title 17 Chapter 1 Section 101:

A “derivative work” is a work based upon one or more preexisting works, 
such as a translation, musical arrangement, dramatization, 
fictionalization, motion picture version, sound recording, art 
reproduction, abridgment, condensation, or any other form in which a 
work may be recast, transformed, or adapted. A work consisting of 
editorial revisions, annotations, elaborations, or other modifications, 
which, as a whole, represent an original work of authorship, is a 
“derivative work”.

No mention of either of your two conditions.  You _might_ be able to 
argue that they use your headers to compile their driver, so that 
violates your copyright, but they are free to develop their own 
compatible headers to produce compatible binaries which are in no way 
derived from the Linux kernel.  See Wine's win32 compatible headers and 
libraries for examples of this.


>> This is why it is legal to reverse engineer a binary driver to gain an
>> understanding of how the hardware operates, publish that information,
>> and then use that information to create new software to operate that
>> hardware.
> 
> No, you are referring to a restriction on the limitations in software
> licenses which is separate from copyright. Copyright law does not talk
> about interoperability at all. And even the applicability of the
> restriction to which you refer is jurisdiction-dependant as well as
> context-dependant (see the DMCA).
> 

No, licenses do not enter the picture at all, I am referring to the Fair 
Use clause of the copyright act and the body of precedence built on it. 
Title 17 specifically states that the study of a copyright work falls 
under fair use and therefore, is not an infringement.  Reverse 
engineering is a form of study, and thus is protected as fair use.

Copyright does not apply to a _process_ ( that's patents ), only to the 
particular expression, therefore, it has been well established in the 
courts that it is not infringing to create a new unique expression that 
serves the same purpose and engages in the same process.  This is why it 
is legal to reverse engineer a windows binary only driver to figure out 
the hardware interface, and write your own driver for linux.

Re: [future of drivers?] a proposal for binary drivers.

[Dave Neuer]

On 3/9/06, Phillip Susi <psusi@cfl.rr.com> wrote:
>
> That is correct, the final decision is up to a Judge, but we aren't in
> court here, we're discussing it, and so I'm making an argument that I
> believe any sane judge would side with.
>
<snip>
>
> A "derivative work" is a work based upon one or more preexisting works,
> such as a translation, musical arrangement, dramatization,
> fictionalization, motion picture version, sound recording, art
> reproduction, abridgment, condensation, or any other form in which a
> work may be recast, transformed, or adapted. A work consisting of
> editorial revisions, annotations, elaborations, or other modifications,
> which, as a whole, represent an original work of authorship, is a
> "derivative work".
>
> No mention of either of your two conditions.

A "work based on one or more preexisting works [in] any other form in
which a work may be recast, transformed or adapted" sure sounds like
it fits someone compiling software with my symbols in it to me.
Elaboration sure sounds like it fits program code calling my program
code to me.

> You _might_ be able to
> argue that they use your headers to compile their driver, so that
> violates your copyright, but they are free to develop their own
> compatible headers to produce compatible binaries which are in no way
> derived from the Linux kernel.  See Wine's win32 compatible headers and
> libraries for examples of this.

I'm sorry, I don't think that analysis is correct for software, see
for example: http://community.linux.com/article.pl?sid=02/11/13/117247&tid=87&tid=41&tid=12&tid=42,
and Linus' previous explanations as I pointed out in my reply to
Xavier.

At any rate, at the moment I'm getting paid software, not for legal
analysis which I'm not qualified to give. I'm certainly not getting
paid enough to sue anyone, so unless some other kernel hacker or
company is planning on initiating a lawsuit (to which I'd happily
join) it's fairly moot, and I'll let it drop at that.

ciao,
Dave

Re: [future of drivers?] a proposal for binary drivers.

[Lee Revell]

On Thu, 2006-03-09 at 16:30 -0500, Dave Neuer wrote:
> At any rate, at the moment I'm getting paid software, not for legal
> analysis which I'm not qualified to give. I'm certainly not getting
> paid enough to sue anyone, so unless some other kernel hacker or
> company is planning on initiating a lawsuit (to which I'd happily
> join) it's fairly moot, and I'll let it drop at that. 

I would think a lawsuit hasn't happened because it would be more
productive to work with these vendors to move the parts of the driver
that must remain closed to userspace.

Lee

Re: [future of drivers?] a proposal for binary drivers.

[Phillip Susi]

Dave Neuer wrote:
> A "work based on one or more preexisting works [in] any other form in
> which a work may be recast, transformed or adapted" sure sounds like
> it fits someone compiling software with my symbols in it to me.
> Elaboration sure sounds like it fits program code calling my program
> code to me.
> 

No, because the individual names of functions are not covered by the 
copyright, only the body as a whole ( or significant part ).  That's why 
the first person to write hello.c can't sue everyone who uses printf().

>> You _might_ be able to
>> argue that they use your headers to compile their driver, so that
>> violates your copyright, but they are free to develop their own
>> compatible headers to produce compatible binaries which are in no way
>> derived from the Linux kernel.  See Wine's win32 compatible headers and
>> libraries for examples of this.
> 
> I'm sorry, I don't think that analysis is correct for software, see
> for example: http://community.linux.com/article.pl?sid=02/11/13/117247&tid=87&tid=41&tid=12&tid=42,
> and Linus' previous explanations as I pointed out in my reply to
> Xavier.
> 

The key question is does work A contain substantial parts of work B?  In 
the case of a source library that is compiled and linked into an 
executable, then you can argue that the executable image is a work 
substantially derived from the library.  In the case of linking to a 
shared object however, the binary does not actually contain any of the 
material from the library, so it is not a derived work.

This is why gcc is not infringing on Microsoft's copyrights whenever 
they create a win32 executable image that links to windows' dlls and 
this is why ndiswan and captive NTFS are not infringing on MS's 
copyrights.

In the case of wine, it is not infringing on Microsoft's copyright 
because they wrote their own win32 api headers.  They contain the same 
function names, but that does not make them a derived work.

In the case of ATI's drivers at least, they distribute their own object 
files which they hold the copyright to, and are not derived from the 
linux kernel in any way, and the user must link them with the correct 
objects of kernel code to create the actual loadable module.  At best if 
you could show that the final module contains substantial code from the 
kernel you might argue that it is a derived work, but since ATI only 
distributes their own object code, there's no way you can claim they are 
infringing on your copyright.

RE: [future of drivers?] a proposal for binary drivers.

[David Schwartz]

> The key question is does work A contain substantial parts of work B?
> In  the case of a source library that is compiled and linked into an
> executable, then you can argue that the executable image is a work
> substantially derived from the library.  In the case of linking to a
> shared object however, the binary does not actually contain any of the
> material from the library, so it is not a derived work.

	Not just substantial parts, but substantial *creative* (as opposed to
functional) parts. Furthermore, are those parts combined with the work in a
*creative* way. (Otherwise you have two works in aggregate form, not a
single derived work.)

	I would argue that linking creates aggregate works just as tar does. The
key *legal* issue is whether the combination is mechanical or creative.

> In the case of ATI's drivers at least, they distribute their own object
> files which they hold the copyright to, and are not derived from the
> linux kernel in any way, and the user must link them with the correct
> objects of kernel code to create the actual loadable module.  At best if
> you could show that the final module contains substantial code from the
> kernel you might argue that it is a derived work, but since ATI only
> distributes their own object code, there's no way you can claim they are
> infringing on your copyright.

	I think that argument would fail anyway. It would require you to argue that
making an actual loadable module from the Linux kernel source and its header
files was not the normal use of those header files. I think that argument
would fail the giggle test.

	RedHat, for example, distributes a 'kernel-devel' package whose primary
purpose is to allow you to turn module source code into module executables.
Since you have the right to normal use without agreeing to the GPL (under at
least first sale), you can *create* those binary modules without agreeing to
the GPL. Remember, there is no right under copyright to specifically
restrict the *distribution* of derivative works, it relies upon you needing
the GPL to give you the right to make the derived work in the first place.

	DS

RE: [future of drivers?] a proposal for binary drivers.

[David Schwartz]

> There are no dearth of legal opinions on this matter which differ
> quite radically from your interpretation here, quite a few from
> lawyers. As far as I am concerned (and the GPL too, if my
> interpretation of it is correct), any code is a derived work of my
> code if either a) it directly makes use of symbols in my code or b) it
> cannot execute unless my code executes, such that its distribution
> without my code would be useless.

	You are claiming that you have copyright over *any* code that *does* X.
This is the one specific thing that copyright can *never* give you. You can
only hold copyright over something if there are a large number of ways to do
the *same* *thing*, and you creatively picked one of them.

	Yes, they can use some code other than your code, just as Static Controls
could have used some printers other than Lexmark's printers. See Lexmark v.
Static Controls.

	But you cannot say you own every way to make X do Y. If there is only one
practically possible way to do X, then nobody can hold copyright on it, even
if there are many ways to do Y, and Y is similar to X. It must be the same.
That's the law, and it makes a lot of sense.

	If you want to own *any* way to make X do Y with Z, then you need a
software patent. Copyrights only cover your choice of one of the many
possible ways to make X do Y with Z.

	DS