Subject: Re: Free Software in Embedded Systems (was Re: Studies)
From: James Horton <horton@cft.net>
Date: Tue, 16 Sep 1997 10:18:30 -0400 (EDT)

Unfortunately, no.

Almost all of our work is strictly confidential, but suffice it to say
that approximately 4 billion microprossors are sold each years, as I have
seen in serveral published sources, but I do not remember the names or
dates of those publications. Most of these are embedded systems of some
sort or another.


One project where we developed a specific 'strategic advantage' for a
LARGE vending machine company, was the use of a single embedded cellular
phone for 50 different vending machines. Standard unix utilities were
use for tight synchronization of time, in addition to a watchdog timer, to
ensure that only one embedded board would answer as the cellular phone,
during its predetermined allotted time slots. Software updates, via
the embedded cell phone, is still a challenge..........


One cellular phone number, 50 embedded systems tranmitting and receiving
information.........

Very low cost and featuring many ported standard unix utilities.

The kernal was custom, due to funky memory organization on the DSP.

The DSP was a standard TI low end chip. Most of the code was written
in C.


The one point I would like to make is simple and good news for tech_heads:

Hardware is nearly a perfect commodity. Software, and the technical 
expertise that goes with it, like a real understanding of memory organization
in various processors, compilers, and the unique abilities of DSPs.........

are what's important. Not major vendor labels.............

After all Ascend and US Robotics still use Phil Karn's old KA9Q ppp/dos
code to manage most of their modem banks in service today........

Livingston has at least risen to the next level...........



James



On
Tue, 16 Sep 1997, Bernard Lang wrote:

> --> Russell Nelson <nelson@crynwr.com>
> > I got paid to write a cs89x0 Linux Ethernet driver for a Schlumberger POS ...
> 
>   Does this mean that they use Linux as an embedded system
>   (just trying to make sure I understand correctly)
> 
>   If yes... any public info on this, on other similar cases ???
> 
>  To come back to the original post about a study on the economic role
> of Free Software, that is precisely what I am working on.
> 
>   But gathering data, papers, arguments is a hell of a lot of work.  I
> have been trying to collect them on a web site, but I am missing a lot
> input from actual industrial users. I learned from 2 separate sources
> (wired, and an independant company specialized in system security)
> that corporate users of free software do not want it to be known.
> Ashamed, I guess :-), afraid for their image.
> 
> Thanks to James Horton <horton@cft.net> for bringing the issue of
> embedded system, that is indeed one of the major arguments. Do you
> have examples that could be credibly cited of such use of freeware -
> or of raping of the embedded world by RTOS vendors. Figures?
> 
> 
> --> Michael Tiemann <tiemann@cygnus.com>:
> > I've heard the same thing.  In fact, we're going to be supporting Linux
> > as a standard host platform for embedded development in our next
> > release, and a number of our F500 companies are very happy about that.
> 
>   Any name you can release ?
> 
> ------
> 
> --> Chris Maeda <cmaeda@alum.mit.edu>
> > software.  Free software only seems to work in niches 
> > that are too small to support proprietary vendors. Once 
> > a market gets big enough, the superior economics 
> > (cheaper for users, more profits for vendors) of the 
> > proprietary model always win.
> 
>   Could you elaborate your argument. What makes the proprietary model
> economically better.
> 
> ----------
> 
> --> "Niels P. Mayer" <mayer@netcom.com>
> > Speaking of embedded systems, I've been chartered to investigate using
> > Java as a glue language to support custom-programmed applications in
> > point of sale terminals. Anybody else thinking about embedded java?
> 
>    I know one fellow who has been developing a reduced Java VM for
> that purpose. I am afraid though that the work may be proprietary.
> I will check.
>    Alternatively, there is a very good VM that could be used freely,
> as good (probably better than Java), but it is not for Java but for ML
> (which is probably a better language too, but that is another story).
> 
> 
> Thanks to all for a most interesting discussion.
> 
> Cordialement
> 
>     Bernard Lang
> 
> P.S.
> For the data I collected see
> http://pauillac.inria.fr/~lang/hotlist/free/licence/ and the list of
> pointers at the bottom of that page.
>  >>>>>>>>>>>>>>>>>>>> Contributions are welcome. <<<<<<<<<<<<
> 
> P.P.S. What is the (IMHO brain damaged) Brooks paper doing in the FSB
> archive, some kind of a joke? Granted he proposes to use freely
> redistributable software, but it does not mean free software by any
> stretch of the imagination.
> 
>  Bernard.Lang@inria.fr             ,_  /\o    \o/    Tel  +33 (1) 3963 5644
>  http://pauillac.inria.fr/~lang/  ^^^^^^^^^^^^^^^^^  Fax  +33 (1) 3963 5469
>              INRIA / B.P. 105 / 78153 Le Chesnay CEDEX / France
>           Je n'exprime que mon opinion - I express only my opinion
>                   CAGED BEHIND WINDOWS or FREE WITH LINUX
> 

**> Creative Friendly Technologies              horton@cft.net 813 980 1317 *
* http://cft.net/access.html       I Peter 3:11   http://cft.net/frame.html *
* Let him turn away from evil and do good, let him seek peace and pursue it *