Subject: Re: quality, sharing, etc.
From: Brian Behlendorf <brian@collab.net>
Date: Sun, 27 Aug 2000 17:00:21 -0700 (PDT)

On Sun, 27 Aug 2000, Rich Morin wrote:
> The Linux model should, in theory, propagate all bug fixes to the "main"
> version of a package and then back out to the various distributions.
> The disadvantage of this is that, when a bug shows up, the fix may have
> to wait on the developer of a given application package.  The BSDs, we
> were told, took responsibility for their own bugs, fixing them at once.

This is primarily through something called the "ports" collection in
FreeBSD, something many of us BSD bigots think is quite cool.

> The flip side of the coin, however, is that a bug fix in one BSD may
> never migrate to the others (let alone to the Linux community).  So,
> FreeBSD has tools that work really well for Intel, NetBSD has tools that
> work on _everything_, and OpenBSD has tools that only work for the right
> users (:-).  A given bug fix may propagate, but this is not the rule.
> 
> Does this seem like a reasonable characterization?  Is this a problem
> that needs to be addressed?  Is there a better approach that should be
> investigated?  Inquiring minds need to know...

I think the BSD's all use the same ports collection concept, but all
have their own instantiation of it.  It'd be nice if these were merged,
and I'd hope that was possible despite the different areas of focus of
each BSD.  

More broadly, if all the BSD's could be accomodated, then that might be
extendable to other OS's too.

It's important to think of the ports collection not *just* as a software
metadata tree, but as a *trusted* metadata tree.  People use it to install
software instead of pulling the package's tarball down separately and
reading the INSTALL instructions because it's a lot easier; required
packages are automatically pulled in, any platform-specific patches are
applied, and you know everything will install in a consistant place.  Note
that this has nothing to do with technology, and everything to do with
the discipline of the ports maintainers.  Could platform-specific teams of
ports maintainers all share the same tree?  I would hope so.

	Brian