Re: libc strategy
On Mon, Jul 02, 2001 at 09:11:13PM -0400, Matt Zimmerman wrote:
> On Mon, Jul 02, 2001 at 05:31:18PM -0700, Will Yardley wrote:
> > it seems to me that these are two different things since bsd isn't a hardware
> > type; it's a kernel and operating system. so it seems to me that
> > 'debian-bsd' should be something different from 'debian-sparc' or
> > 'debian-alpha' since it's not using a linux kernel.
> Yes, it will. It will be very much like Debian GNU (aka hurd-i386).
If it becomes an issue, I think the best thing to do would be to label packages
by kernel as well as architecture, so control files might say something like:
If the Hurd ever runs on anything besides i386, this would probably be useful,
> > i am not a huge expert on operating systems, but it seems that using gnu
> > utilities and ext2 and making it more gnu/linux-like is taking away a lot of
> > what makes *bsd so great; not that these are bad, but if people wanted them
> > they'd use gnu/linux :>
> > looking through the archives it seems that this is the never-ending debate...
> Everyone wants something different (BSD/GNU userland, Free/Net/OpenBSD kernel,
> BSD/GNU libc). That is not nearly so much of a problem as the fact that almost
> none of the camps have sat down to do much real work.
> Whoever builds the first prototype gets to choose the direction for the
> project. If nobody wants this badly enough to actually do it, it will probably
> never happen.
You're essentially correct. I'm looking at FreeBSD, and I do plan to spend some
time getting dpkg working, and I'm going to see if I can package the BSD system
as it is. (Or parts of it, anyway.) That would be a start. From there, I plan
to look into getting the Debian glibc running under compat, and getting some
basics like sysvinit working.
> In case you're wondering, I'm subscribed to this list because I would be
> interested in the result, not because I'm motivated to spearhead the project.
> This list seems to be mostly populated by people with a similar view.
> Is there anything particularly spectacular that one would be able to accomplish
> with a Debian/BSD hybrid, something to make the sweat worthwhile for at least a
> few people here? Does it scratch an itch? So far, it doesn't seem to be much
> more than an "It would be nice if..." with some dpkg/apt diffs as a starting
I have specific interests in some of the FreeBSD kernel features that don't
presently exist or don't work well on Linux. For example, NFS on Linux is
definitely not all it could be, and Vinum is supposed to be pretty stable,
while LVM on Linux is still a bit messy, and not well integrated yet.
While I'd like to have the kernel, I don't really want to deal with a yet
another *nix. I already deal with Debian and Solaris (ick) every day, and I
don't really want another completely different system. If it's enough like
Debian, it wouldn't be a big deal. I don't really like the FreeBSD
configuration and packaging. It's essentially the same way I managed Slackware
systems a couple years ago, and there's no way I'm going back.
Also, I think that porting Debian to another kernel would help the Debian
project as a whole, because it's been my experience that porting software
to a new OS or kernel frequently turns up new bugs, especially in programs
that haven't been ported a lot. Porting some of the essential Debian tools
off of Linux is good for them.