Social Contract: Practical Implications
I have waited about a week to calm down to write this note. I hope it
reflects that wait. For those who's patience has thinned during this
long conversation, you can skip the Preamble, and go straight to the
questions.
Preamble:
I am a fairly ardent user/supporter of Debian. From a practical pov,
apt-get is a great accomplishment. I have become completely used to the
spartan installer and even run a small (~64 node) Beowulf on it, using
only MPICH for parallelism.
I have also become accustomed to the tension between some (possibly
many) of the debian developers who would like a very strict
interpretation of "free" and a strict adherence to this interpretation,
and others, including many users and possibly developers, who see
development of a strictly free distribution
as an important, even guiding, goal, but who have pragmatic motivations
including performance and timeliness of release that would lead to, at
the very least, a much longer time-line for strict interpretation.
While, as a user, I believe I'm not in a position to vote on the general
resolutions, I would like to understand their implications. I will even
be so bold as to suggest a path.
Questions:
1. I was under the impression that firmware leads to better performance.
Is this true?
2. If 1. is true, then will removing the bits of firmware from the kernel
lead to degraded performance of, for example, graphics cards?
3. Are there other features, besides firmware, whose performance will be
effected adversely by the strict adherence to "free" envisioned by the
social contract?
Leading Questions:
1. If there is a degradation of performance, is anyone thinking of
establishing an alternate path through the archives, using apt-get,
where, for example, kernel packages with firmware would be placed in
non-free and would be available by configuring the apt-get
appropriately?
2. Is there a movement to establish a relaxed Debian distribution, with
an independent archive that is maintained to be as close as possible to
the ideal Debian archive, but where performance is given equal weight to
strict interpretation of "free"?
In my work place, a government laboratory, we have had discussions
about Debian and its relationship to the other
distributions. The prospect of waiting another year until Sarge has
protection of the security site is very discouraging and has already
led to many abandoning Debian for Suse. I have still not been able to
contemplate life without apt-get, but if new versions of the kernel lead
to degraded performance, I will not have a lot of choice.
This discussion of "free" is a more devisive version of one that I remember
from a few years ago. My POV has not changed much. If truly free software is
even close in performance to commercial versions, there would simply be
no argument about its use. As an exmaple, I point to mozilla. During the
last conversation, I pointed out that I would use it when it was ready
for prime time. It's now my default browser. I would suggest that this
would be a guiding principle. It would mean that Debian would never be
strictly free, because innovations happen in both the free and
commercial software communities, and development of free alternatives to
copyrighted software takes time. Ignoring these innovations would lead
to a strictly free, strictly out of date distribution that would not
serve the greater linux community. I would argue that it would not even
serve the Debian community.
Art Edwards
Reply to: