The proposed GR: catch-22
It seems to me that this proposal, like the similar one by Wichert
about a year ago, seems to be bolstered by its supporters with two
mutually exclusive arguments:
1. It will have minimal effects since people will still be able to use
non-free software.
2. It will have far-reaching effects since Debian (the corporate
entity) will expend fewer resources in support of people wanting to
use non-free software.
As others have pointed out (most notably, Michael Alan Dorman), #1
isn't really the case since the most likely result of a dropping of
non-free will be a "grey zone" of debs from random sources, a la RPM
contrib packages. We're already seeing serious compatibility problems
with < 5 serious producers of .debs (Corel, Stormix, TDYC, Helix Code,
plus ourselves*); I suspect the "apt sources list" that Stephane puts
together shows even more.
#2 seems to be the more likely outcome, but it contradicts our promise
to be guided by the needs of our users. Maybe once we have popcon
results showing nobody ever uses stuff in non-free, we can consider
this proposal, but I'm not seeing them yet. (Instead, I'm seeing
packages like netscape, fragmented though they are by versions, being
stuck on the first CD of a "all sections" potato set.)
I'm really rather surprised that people who don't do anything with
non-free packages are burdened by their presence. The incremental
cost of maintaining the BTS and other tools (lintian reports, etc.)
for a few dozen packages can't be all that high. I submit that it's
people who package non-free software that carry the bulk of the
effort, and they're volunteers by definition (unless you count -qa).
I guess I'm missing the real impetus behind this proposal...
Chris
* This is not a slam against these other .deb packagers; it's just the
inevitable result of development from a divergent tree that isn't
monitored for dependency consistency with Debian proper.
--
Chris Lawrence <lawrencc@debian.org>
Reply to: