[Mageia-dev] cinelerra/audiokonverter/arista (war Re: rehashing the faac issue)

Wolfgang Bornath molch.b at googlemail.com
Thu Nov 1 23:44:52 CET 2012


2012/11/1 Christian Lohmaier <lohmaier+mageia at googlemail.com>:
> Hi Wolfgang, *,
>
> On Thu, Nov 1, 2012 at 1:56 PM, Wolfgang Bornath <molch.b at googlemail.com> wrote:
>> [...]
>> After reading all arguments again I must confess that I changed my
>> opinion: Being consequent and following our road we need a
>> /tainted-free and a /tainted-nonfree branch.
>
> I still think this would be a very user-*un*friendly way to handle it.
>
> You cannot put packages that itself is "free", but depends on
> "tainted&nonfree" packages into the "free" repo, since the core repos
> need to be self-contained.
>
> The natural approach would then be to put that package into the same
> repo as the packages that fulfill the requirement.
>
> If now you have two tainted repositories, one "free" and one
> "non-free", you would put the "free with dependencies" package into
> the tainted-free repo. But it itself isn't tainted, so actually
> wouldn't belong in there. But even when you have tainted-free enabled,
> and you try to install the package, you need to enable the
> tainted-nonfree one.
> Or you put it into tainted-nonfree to keep the repos more
> self-contained, but then the distinction is rendered useless, as
> packages end up there for completely different reasons.
>
> And if you have to enable both tainted variants anyway, there is no
> point in having them separate in the first place other than to please
> some bureaucratic nitpicking.
>
> So by creating a tainted-nonfree repo that only a handful of packages
> actually belong into anyway, you create a situation of non-satisfiable
> dependencies that make the distinction pointless from a user-POV.
>
> Much better would then be to create an "ugly" repo (in the spirit of
> gstreamer) that contains the "doesn't fit into the other repos" stuff.
>
> AFAIK only multimedia related stuff falls into tainted-nonfree. And it
> is either you want to use it or not. If you want to use it, the user
> doesn't care whether it is free or nonfree (by whatever definition,
> there isn't the one-and-only definition that everyone agrees with
> anyway). You only might care about whether it is tainted or not (and
> many don't give a damn about software patents even where they could
> apply in theory). It is not like you have a chance in most of these
> cases. Either you use a "100% free but sucks" implementation, or you
> use the "tainted and possibly "nonfree", but working just fine" one.

Taking what you wrote into account there is no chance for a solution
which also honors the division between free, non-free and tainted, at
least non of the propositions in this thread would work. Because as
you described the problem is not with each software but rather with
its requirements (free software needing a non-free or tainted library
to run or offer all its features).

So what can we do? What do users want in such cases? First they do not
want crippled packages. If they decide to have mplayer they want the
whole thing, not a free variant which lacks features. They'd rather
not want mplayer at all.

Answering to this there may be a rather pragmatic but also drastic and
even provocative solution: Stop offering software in core which need
non-free and/or tainted libraries to give all its features. This
software should not be offered in 2 or even 3 variants but only in one
full version, residing in an "ugly" (or whatever you want to call it)
repository.
Yes, there may be people who would like to use mplayer without
non-free or tainted libraries - but are there really? To my experience
they either use the full blown package or another software.

This solution may not hold against theoretical arguments but I think
it's pretty much the practical way, for users as for packagers. And it
will motivate people to search for free solutions where possible.

-- 
wobo


More information about the Mageia-dev mailing list