Re: ftp.freelinux.org

[Jim Pick <jim@jimpick.com>]

> In addition, we're working on central-source compilation
> (Klee said he'd work on it as part of our security team), so you didn't
> need an organization for that. So I'm still confused about what you
> wanted to achieve by creating a separate organization.

I'd like to echo Bruce's sentiment.  dpkg development seems back on track
(with Klee and others), and it should be possible to work towards using
CVS for the core packages (and others), on a selective or global basis.

I agree the source format is broken (I was quite active in debian-devel
on this topic), but there's lots of opportunity to fix it.  If we added
source dependencies, that would make bootstrapping to different
architectures quite easy -- and I'd argue that it's probably easier to
have multiple independent source trees that have to be ported rather than
having one large, ill-defined one based upon a CVS repository.  (I'm 
currently trying to get dpkg going on Cygwin32, so I understand the
issues, I think)

There's definitely merit in using CVS and having a single source tree that
all developers can contribute to (a la FreeBSD) -- but there's also merit in 
having independent packages too.  The 'independent packages' approach scales
quite nicely (just look at the number of packages we have).  

The FreeBSD approach works for them, since they wrote their own kernel -- 
I think they wrote everything in their base system -- so they have 
complete control when it comes to testing and release control.

With Linux, the kernel effort is a separate effort.  So are most of the
packages that make up the base system.  As a result, Debian has evolved to
be a collection of independent efforts cobbled together by volunteers.  It
seems to be working.  The 'base' fully-tested Debian distribution seems pretty 
solid (as of 1.3).  The advantage of having a base system made up of 
loosely co-ordinated efforts is that we can 'pick + choose'.  

There's advantages to the package-based approach.  If for some reason, Linux 
kernel development takes a left turn, we could switch to GNU Hurd.  We could 
concievably host the Debian environment on top of cygwin32 (for the 
Win95/WinNT crowd).  Developers tend to work harder and get more accomplished
if they work on one thing and are responsible for it (thanks to RMS for
pointing this out to me).

My point is -- I'm using Debian because it's technically the best thing out
there -- and it's the largest collection of "free software" out there -- not 
FreeBSD, not Red Hat.  Debian's also got a lot of momentum, which means it 
will get better faster than anything else.  

What does Linus say?   "World Domination. Fast"  :-)

In summary:

Obviously (to me anyways), Mike + Dominik & others, are looking for a
totally different organization.  We know there were personality conficts.
I don't see a clear technical direction that differs significantly from
what Debian wants to do.

I also don't think it's very constructive basing a new effort on animosity
towards a selected group of individuals (us Debian).

However, in order to be successful (and fracture the free software community
even more), they will have to come up with something that is technically
better than Debian - but with 200 less programmers (including some really
top-notch ones) - and before we do the same thing.

So,

for the "FreeLinux" folks:

Wouldn't it be better if you could kiss + makeup, add Bruce to your killfile 
(he hasn't been particularily bossy lately anyways), and consider re-joining 
the Debian project.  Instead of building your own cvsup setup based on the 
FreeBSD stuff, why not work on building one that would work hand-in-hand with 
an improved dpkg and source packaging scheme?  

That would be cool.  :-)

Cheers,

 - Jim

Attachment: pgp47G9l0JG6s.pgp
Description: PGP signature