[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[bluetooth-dev] RE: free-unix device drivers for digianswer pc cards

> > Here at Digianswer, we would like to provide hardware support for our
> > devices on Linux. As our hardware interface is strictly confidential,
> > driver support will be on a binary-only basis. The source-code will,
> > under no circumstances, be freely available.

Sorry for the delayed response, I've been busy with CeBit, WinHEC and the
resultant backlog of mail.

> My first question is why was it that Digianswer decided to make the
> hardware interface "strictly confidential"?

Basically so that other stack vendors cannot exploit our hardware (which is
currently the only commercially available hardware) to support their stacks.
We want to establish a market presence for BOTH our hardware and our
software. This is mainly because our software is markettable seperately from
our hardware. Being the only software stack on the market is a huge
advantage when other hardware vendors come to market and need a stack to
support their hardware.

> Who can reverse this decision?

Currently noboby.

> Is there some problem with the hardware interface that demands this
> secrecy? Can Digianswer give me written assurance that no security
> or reliability problems will arise from my use of their confidential
> drivers?  How can Digianswer assure me that all possible performance
> gains achievable at the hardware interface level have been made?

Nobody can and nobody will give you written assurances about the stability
or security of their products. Your point is that if the driver software was
open then the miriad programmers out there could inspect and optimize it for
us. Firstly, I think the majority of those programmers are only concerned
about getting their Bluetooth software working - they'll only look in the
driver code if there's a problem. Those remaining programmers who do look at
the code for the fun of it and are willing to dedicate time to optimizing it
could possibly increase the stability and security of the stack - but this
option is not open to us because our driver is closed.

> Why is Digianswer paradoxically embracing the open source model with the
> BTSWS effort yet ignoring the open source model w.r.t. free operating
> system device driver support? Is it not true that Digianswer is just
> cynically using BTSWS for public relation reasons? How can Digianswer
> claim to support one public, collaborative effort, yet reject and indeed
> inhibit another within exactly the same model?

We don't go open source until 2004. At that point EVERYTHING will be open
source INCLUDING the device drivers. At this point, those interested will be
able to deduce the interface from the driver code. We are not using open
source as a public relations stunt. We ARE going open source in 2004 but NOT

> If/when Digianswer do release a closed driver for Linux, won't an
> avoidable cost of quality be passed on to the customers? i.e. Digianswer
> must choose between an expensive, high-quality closed driver, and/or
> a cheap, low quality driver. Both alternatives are cost-ineffective
> when compared to a device driver developed by open source enthusiasts
> from a completely open hardware interface. How can Digianswer remain
> cost-competitive by keeping their hardware interface confidential?

Digianswer are willing to pay the extra cost to produce a driver that is
developed and tested by the correctly qualified and experienced people in a
timely fashion. Your cost comparison is quite irrelevant, in my opinion.

> Won't customers who have purchased Digianswer PCMCIA devices lose all
> support for the accompanying software when Digianswer disappears? Isn't
> this a kind of cross-vendor lock-in? E.g. a future version of Linux or MS
> Windows may not be compatible with the last released Digianswer drivers.
> What risk or cost will customers have to bear when Digianswer inhibits
> them from changing to different or newer versions of an operating
> system?

Again, open-source in 2004.

> How can a customer's investment in purchased hardware be justified when
> the information required to make general use of that hardware is kept
> from them. Isn't Digianswer dictating the scope of the customer's
> ability to use their purchase?

So you're saying that anybody who buys a piece of hardware that is supported
by a closed driver is looking for trouble? This is ridiculous! The average
users decision to buy a piece of hardware is based on it's functionality not
whether or not he can develop code for it himself! If you don't WANT to buy
the Digianswer card because there isn't an open-source driver for it, then

> And a question that someone else may be able to answer: Why didn't the
> Bluetooth specs include a basic interface for PCMCIA? Why was it
> explicitly left as vendor-determined? Isn't this a blatent endorsement of
> non-interoperability that is wholly contraditory to the purpose and intent
> of publicly released standards?

I have to agree with you here. It would've been a good idea (and more
consistent with the other hardware interfaces) to develop a common PCMCIA
standard. I really don't know why this wasn't done.

> Looking forward to any answer, although I expect to be ignored.

Actually, I very nearly did ignore this email. There is a large degree of
animosity in your email and a certain degree of naivety as regards the
commercial product market as a whole. As you weren't expecting an answer,
one could hypothesize that you are merely venting your anger at not being
able to get at the source code, whilst scoring points amongst those
open-source sympathizers on the list.

I have personally decided to produce driver support for our cards under
Linux but I'm doing it under the realistic market-driven conditions that a
profit-making concern dictates. My interest, and that of my company, is to
drive Bluetooth forward, not Linux. If this is the kind of reaction I can
expect from the Linux community, then I might as well just save myself the
headache and quit now - this would certainly not affect Digianswers profits
or strategy in any way. Which would you prefer: a closed driver for the only
widely available hardware or no driver at all?

I am not trying to start a holy war here and I certainly don't have the time
to partake in one but Digianswer is a company - it exists to make money and
has a well defined strategy for doing this. Giving away source code to the
general public at this point is not part of that strategy and would on the
contrary negatively affect that strategy.

Antony C. Roberts,
Digianswer A/S.

> David Leonard                           David.Leonard@xxxxxxx.au
> CRC For Distributed Systems Technology  Room:78-632  Ph:+61 7 336 58358
> The University of Queensland            http://www.dstc.edu.au/
> B73CD65FBEF4C089B79A8EBADF1A932F13E=
> A0FC8
> [Opinions expressed or implied above are mine, and not of my employer's.]
> ---------- Forwarded message ----------
> Date: Tue, 13 Mar 2001 13:19:06 +0100
> From: "[iso-8859-1] Mattias =C5gren" <mattias.agren@xxxxxxx.com>
> To: bluetooth-dev@xxxxxxx.com
> Cc: "Bluetooth (tm) News and Discussion ** VOTE-->"
> <bluetooth@xxxxxxx.com>=
> ,
>      "'bluetooth@xxxxxxx.com>
> Subject: RE: [Bluetooth/PAN] Bluetooth Card
> > Hi,
> > I search cards for Bluetooth's connection between two Pcs on Linux.
> > I use Axis stack and i search compliant cards.
> > If my understanding is good i can't use PCMCIA card.
> > Have you find solutions for use it ?
> >
> > =09Tanks for your help
> >         =09        Tony
> >
> not yet... but take a look on this article on digianswers site
> (or below) :
> http://btsws.digianswer.com/btsws/articles/info0003.asp
> =3D=3D=3D
> Linux Driver Support - Help us to help you
> By Antony C. Roberts, Digianswer
> 08/03/2001
> Here at Digianswer, we would like to provide hardware support for
> our devic=
> es on Linux. As our hardware interface is strictly confidential,
> driver sup=
> port will be on a binary-only basis. The source-code will, under
> no circums=
> tances, be freely available.
> The goal is to enable customers to run on open source stacks such
> as those =
> sponsored by Axis and IBM (Blue Drekar).
> I am currently personally undertaking this goal on a low-priority
> basis and=
>  am hereby requesting help. I am an experienced Windows and
> Solaris device =
> driver developer but I think I can more effectively achieve my
> goal with he=
> lp from experienced Linux PCMCIA driver developers, especially as
> the kind =
> of time I can dedicate to this project is very limited.
> For initial development purposes, the target platform is RedHat
> 7.0 and the=
>  target itself is a PCMCIA client driver.
> I have already started on the development and have made a
> skeleton driver t=
> hat currently does nothing more than attach.
> I would like to start by appleaing for pointers to relevant
> resources, such=
>  as good documentation on the Linux PCMCIA initialization process
> and sourc=
> e code to similar drivers - the Digianswer PCMCIA devices use the
> Analog De=
> vices ADSP 1801, so the ideal goal would be a driver for a device
> which als=
> o uses this DSP.
> If you can help, please contact me at acr@xxxxxxx.com, or post
> to dga.bt=
> sws.linux at news://news.digianswer.com

To unsubscribe from this list: send the line "unsubscribe bluetooth-dev" in
the body of a message to majordomo@xxxxxxx.com