On Tue, Jan 06, 2004 at 09:12:00PM -0500, Dale E Martin wrote:
> Benefits
> --------
> - Perceived philosophical benefit in supporting only DFSG compliant
> software as a project.
Presumably you should list "Perceived philosophical costs in not
supporting as a project all the software we legally can" in the next
section.
The easiest way to deal with perceptions is either to ignore or remedy
them when they're mistaken, or to accept them and work out if there's a
better alternative when they're not. In neither case are they anything
more than a cue that it might be worth considering what's going on here,
though.
> - Small reduction in bandwidth, disk space, and cpu cycles on project's
> servers.
The key point is the reductions are so small they're likely to get lost
in the noise. If we drop non-free now, and include the AMD64 architecture
next week, our bandwidth usage, disk space usage, and cpu usage will
all be much higher in a week than they are now. IMO, this isn't worth
considering.
You are, though, missing a couple of potential benefits. If we drop
contrib too, it becomes substantially easier to maintain a bunch of
infrastructure, because we've suddenly only got one place to look: main
(non-US is almost entirely for non-free stuff as is). Dropping contrib
also means we can fully support everything we distribute, although I'm not
sure what practical benefits that might have. Some benefits in handling
security updates might be plausible.
> Costs
> -----
> - Alienation of users who find changing their sources.list to be
> problematic.
Uh. How about not being quite so dismissive, and saying "alienation of users
who (a) can no longer find debs for non-free software they use, (b) have to
manage tens of independent apt sources of varying quality instead of one"?
The problem isn't editing the file.
> - WTF to do with contrib?
That's not a cost, it's an issue that needs to be decided, but hasn't.
If you want to add a benefit/cost pair to the proposal on the table, though,
you've got:
Benefit: removes conflict between "Debian will be 100% free" and
"Debian supports non-free and contrib"; removes confusion about what
the "Debian distribution" is, exactly.
Cost: creates conflict between maintenances of contrib, and claim that
"we will never make the system depend on an item of non-free software".
> Feel free to add to either list. I almost put "loss of functionality such
> as support for certain hardware, japanese pdf readers" but technically
> Debian does not contain those now.
No, the "Debian GNU/Linux Distribution" does not contain them now --
that's what main is. Nevertheless, Debian does offer support for those
things through it's non-free section, and we would lose that support.
> No, my point was that there are Debian developers already willing to
> maintain parallel infrastructure on or above the level that would be
> required by nonfree.org, so on the surface this doesn't seem like a
> showstopper.
Perhaps. I think you're underestimating the problems though -- most
external apt sources don't bother with most of our infrastructure; they
don't bother validating uploads (since only one person can upload), they
don't have good mirroring or mailing lists, they tend not to have BTSes,
they tend not to have any special security support, they tend not to have
a process for letting independent people add software to their archive
(ie, new-maintainers). Mostly this isn't a problem, because the software
they're maintaining is intended to get into the Debian archive sooner
or later, and doesn't really need it's own long term support structure.
Certainly, it's _possible_ for people to recreate this. One way or another,
all the software and procedures we're using are available for people to
reuse.
The question, though, is whether whatever _actually happens_ will be as
good as what Debian currently offers our users -- a single site, with lots
of infrastructure is a lot more useful than a hundred poorly supported sites
to try to find then look through.
And the question is also whether whatever actually happens will cost as
little time and effort as our support of non-free currently does. If it
in fact takes more time or resources or effort than reusing Debian's
existing infrastructure, then it's highly likely that that time and
resources will come at a cost to free software; ie, that it's time and
resources that would otherwise have been spent doing stuff that would
be useful for free software.
Both those are important questions: there's little we care *more* about
than how our users are affected, and maximising the resources available
to the free software community of which we're a part.
Cheers,
aj
--
Anthony Towns <aj@humbug.org.au> <http://azure.humbug.org.au/~aj/>
I don't speak for anyone save myself. GPG signed mail preferred.
Linux.conf.au 2004 -- Because we can.
http://conf.linux.org.au/ -- Jan 12-17, 2004
Attachment:
signature.asc
Description: Digital signature