[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: [rms@santafe.edu: Hurd packager]



Raul Miller <rdm@test.legislate.com> writes:

> I've created the mailing list debian-hurd@kato.legislate.com, and have
> created a (rather bare) web site for its archive:
> http://kato.legislate.com:82/debian/hurd/
> 
> I've subscribed gord (and myself) to the list. If anyone wants to
> subscribe, send email to debian-hurd-subscribe@kato.legislate.com
> (you can also retrieve copies of past messages by mail -- you'll get
> instructions when you subscribe).

Sounds cool.  I've signed up too.
 
> I'll be happy to move the list (and its archive) over to debian's
> primary list server, if that's later deemed the right thing.
> 
> For now, there's one real big issue to tackle: what needs to happen?
> 
> Is this like a port to a new architecture?
> Is this like a new kernel package?
> Are we trying to create new boot disks?
> Something else?

I just grabbed the development Hurd boot/root disks (3 floppies), and
tried it out here.  It worked on my Pentium, but not on my 386 (which
would be where I'd install it).

It doesn't look like it would be too hard to get dpkg going on it.
Most of the other GNU tools are already there.

The GNU Hurd web pages are awful - all the information is out of date,
and there isn't even a mailing list archive.  Blech.

What would make things really easy is if the GNU Hurd was being
designed so that it would have binary compatibility with Linux.  That
way, we could just use our normal Linux packages, and just have a
different base system under it.  (Theoretically, we could do the same
thing with the FreeBSD kernel, because it has Linux emulation)

I have no clue as to whether or not that is a design goal.  Perhaps
there are good reasons not to have binary compatibility with Linux.

(Remember the 86open project?  http://www.telly.org/86open/ )

Of course, what makes the Hurd interesting is that it can have
multiple personalities, including non-Unixy ones.  I doubt it would
ever beat Linux as a Unix kernel, because Linux is just a Unix kernel,
and highly optimized at that.  But there are already a lot of cool
kernel ideas floating around that are never going to go into Linux,
because they break Unix traditions.  So the Hurd may acquire some
interesting, new capabilities - and there will be non-portable
applications that use these unique capabilities.  That's why I think
it might be a neato alternative kernel for Debian.  (Hosting Debian on
Cygnus GNU/Win32 would be sortof cool too)

Cheers,

 - Jim

Attachment: pgpnKI32UrFiM.pgp
Description: PGP signature


Reply to: