Hello all,
This seems late to reply in response to this thread, but I thought
that something is worth stating explicitly, as I didn't see it
anywhere (I could have easily missed it, it's a long thread in a list
full of long threads lately).
As others have pointed out, the basic tools are there to custom-build
packages from source. There's apt-source and apt-build, as well as
apt-get source itself. We don't have some of the other features of
gentoo like slots that people have brought up, but that's not an
impossibility for the future.
The burden and inequality, as well as the questionable benefits of
optimized binaries has been discussed at length, and I think that
gentoo itself shows that optimized binaries don't have to be
distributed. If people really want them, they will use a system like
gentoo and compile it themselves. Debian does not have to build them
for the users.
That said, as for a large-scale mechanism to build optimized
binaries, if people really want it in Debian, I suggest that they
simply put up the code for it. Take apt-source or apt-build and start
modifying them. Take dselect, aptitude, or synaptic and provide an
interface to apt-source or apt-build. Start testing d-i and start
writing code that will help a user build a system from scratch if
they really want it (help get d-i in a fully functional state first
though!)
So if we, the Debian community Andrew refers to, really want this,
then "we" should implement it the same way everything else around
here gets implemented. People do the work that they want to do. If
people want a full ports system in Debian, get coding. Rather than
worrying about the people who run gentoo instead of Debian (or Redhat
or SuSE or whatever instead of Debian for that matter) worry about
making Debian the best system that it can be, and that means writing
code.
- David Nusinow
[EMAIL PROTECTED]
