Frank Griffin a écrit :
On 01/09/2012 06:38 PM, Anssi Hannula wrote:
On 10.01.2012 01:30, David Walser wrote:
That's the absolute last thing I want to see happen. It's one of
the reasons Fedora and others that do that are not viable options
for a lot
of non-technical users, and it just makes it so you have to jump
through a lot of extra hoops just to have a reasonably working system
I'm absolutely fine with either moving codecs to core or tainted, as
long as we are at least somewhat consistent in what is in core and what
is in tainted.
I agree with both of you, but the problem is the difference between
non-free and tainted. Unfortunately, both are required for a
"reasonably working system". But in the case of nonfree, it's just a
matter of us internally getting past the "purity of the media" issue.
In the case of tainted, there are legal issues.
Patent claims are potential legal issues, since among other things, most
claims are invalidated when (or before) they reach the courts. So it
makes much more sense to wait for real legal patent threats. Agreed, it
does make sense to put packages in "tainted" when real threats arise.
Such cases will impact much bigger and richer targets before us, so we
will have plenty of time to react.
I think that the ultimate solution is going to have to be integrated
support in the installer allowing the user to hook up either nonfree,
tainted, or both, just by answering a simple prompt. The mirror
database probably ought to (if it doesn't already) keep track of which
mirrors supply nonfree and tainted (or better still, check at install
time), and customize mirror selection based on the user's response.
Good idea.
Note that only "tainted" is optional on official mirrors. (Yet they all
seem to carry "tainted", including in the apparently "patent-threatened"
countries such as the U.S.)
For non-network installs, I don't see a way around separate ISOs, at
least for tainted.
Another reason why we should limit what goes into tainted. Since many
users don't have a connexion reliable enough for network install.
Our goal should be that it "just works out of the box", for as many
users as possible.
Without unduly compromising our commitment to open source, of course.
As for separate ISO's, if we go that route, we should arrive at some
solution that minimises the work involved (including testing) to produce
them. Such as maybe producing a DVD with enough non-free firmware and
drivers to get almost everyone's hardware working properly, and then
excluding the few non-free packages to produce a totally free DVD for
the purists.
(The non-free kernel-firmware package is only about 20M, for example.)
--
André
