On Fri, Oct 29, 2004 at 08:58:11AM -0400, Michael Poole wrote:
> Someone might implement a good free synthesis and place-and-route
> toolchain; assuming the major EDA and FPGA vendors do not sue the
> developers for patent infringement, it might only take 5 or 10 years
> to make it support a reasonable fraction of FPGAs out there.
>
> Firmware for processors is obviously a different case, but there is
> just as much pressure to open or reimplement that code when the
> firmware is in non-free as when both the firmware is non-free and the
> driver is in contrib.
No, there isn't; if the driver is in main, the implication is that it is
functional on its own--it isn't. People expect things that require non-
free parts to be in contrib; that's why it exists. ("It's hard!" doesn't
make it free.)
> Thus this discussion. Perhaps obviously.
Lobby for change would go on d-project, wouldn't it? (D-legal is mostly
about DFSG interpretation--free vs. non-free, not free vs. contrib, so
I guess it's been off-topic from the beginning.)
> Non-free servers all require non-free pieces of data: the server.
> Without that, the client is not complete.
>
> You think how the server or device gets non-free data is important. I
> think the low-level interface is more important.
Non-free servers are outside Debian, just as hardware is outside Debian.
If they require us (the client) to store non-free data and send it to
them, *that* is inside Debian--the user's machine needs to have a copy
of that data.
We're beginning to talk in a pretty tight circle, and it doesn't seem like
we're getting any closer to the core of our disagreement.
--
Glenn Maynard
