Ubuntu is not free.

[Mario Vukelic]

On Wed, 2006-07-12 at 23:49 -0700, Matthew Kuiken wrote:
> The firmware that runs in a card's microcontroller tells amazing amounts 
> of information about how the firmware and hardware are partitioned, 
> which algorithms are hard-coded versus done in updates, what 
> microcontroller is being used, etc.

I trust your knowledge as an embedded systems designer, but to my
layman's eye this information does not look earth-shattering. I'd figure
anyone intent on copying your hardware and actually being able to build
a copy on an industrial level has other ways to find these things out.

> As an embedded system designer myself (both hardware and firmware), the 
> thought of releasing this kind of code to the public is just mind 
> boggling.  

The same could have been said (and was) about the crazy idea to release
the code to an actual working industrial-strength operating system. I'm
sure MS equally finds the idea of releasing the Office source
mind-boggling.

> I have in the past refused to deliver this code to other 
> devisions _of my own company_ in order to maintain a proper development 
> path that did not involve unsupportable patches.

I think bad coding practices in your company are a whole different
matter. The same argument could be made for not releasing the source to
linux, because then some stupid or time-pressured developer could write
code that depends on data structures that are internal to the kernel.

>  I even end up moving 
> the line between hardware and software around during the development 
> cycle.  Sometimes I will move a software function to hardware if the 
> processing warrants it.  More often, I will move a feature from hardware 
> to software, because the chip doesn't work quite right.

I agree in this case. In an early posting in this thread I said that I
personally find closed firmware not _that problematic, since it can
somewhat be considered a part of the hardware, and could nearly equally
well not be a binary blob in the kernel but live in a chip on the
hardware.

> All of that said, my systems always include an external API, usually 
> controllable through a serial port, which is used to control the 
> device.  This API is freely available to anyone who is buying our 
> product, internal or external to the company.

A good thing I guess, except if i want to do something with my hardware
you didn't think of.

> One last point, even if Intel provided said firmware code, since it is 
> for a (most likely proprietary) microcontroller, are you even certain 
> that you could compile it...  As I think about it, the firmware could 
> even be the programming instructions for an FPGA.  If so, I don't know 
> of any free software out there to compile VHDL/Verilog for anything 
> bigger than a million gates.  I invite you to try to keep up with the 
> design changes from Xilinx/Altera if you'd like to try.

See "part of the hardware" above. However, one can use a closed source
VHDL/Verilog compiler to do the stuff one needs to do, and without
violating any free software principles. GNU was written using
proprietary tools, and the FSF has a text on its site explaining when
they consider using proprietary tools ok.