Re: kernel-source == Linux or Hurd or ???

2003-09-26 Thread Eduard Bloch
#include 
* martin f krafft [Mon, Sep 22 2003, 08:03:18PM]:

> This is a good point. Debian makes an effort to be kernel
> independent, so why does the kernel-source install Linux?
> 
> I think we should rename to linux-kernel-source, linux-kernel-image
> and so on...

Good point. The idea is not new and there have been discussions about
new kernel naming strategy (including module packages names,
installation method etc. pp.). If you wish to participiate that
discussion, add your comments to:

http://debian.linuxwiki.de/DebianKernel/Plan

But be warned, this is only planning for the future development
direction now.

MfG,
Eduard.
-- 
Takt ist, wenn man's genau betrachtet, eine Art von Hellseherei.
-- S. O. Jewett




Re: kernel-source == Linux or Hurd or ???

2003-09-24 Thread Joel Baker
On Wed, Sep 24, 2003 at 01:56:09PM -0500, Ryan Underwood wrote:
> 
> Hi,
> 
> On Mon, Sep 22, 2003 at 08:03:18PM +0200, martin f krafft wrote:
> > 
> > This is a good point. Debian makes an effort to be kernel
> > independent, so why does the kernel-source install Linux?
> > 
> > I think we should rename to linux-kernel-source, linux-kernel-image
> > and so on...
> 
> I very much agree with this sentiment.  However, what about system-level
> utilities and essential packages?  How does a Debian *BSD system differ from
> a Debian GNU system (or for that matter, a Debian GNU/*BSD system), and
> how should the dependencies between "bare-metal" packages and the kernel
> (whether it be linux-kernel-source, freebsd-kernel-source, etc) be
> constructed?

The differences depend a lot on whether you view them as "what package is
it in and how is it provided" (significant) or "what the user sees" (as
minor as we can manage).

> Maybe this is silly, but perhaps the "arch" portion of the apt sources
> could also be fine-tuned to include the kernel type.  (I guess similar
> to the unique machine strings from config.sub and friends).  A
> linux-gnu-i386 distribution, a freebsd-gnu-i386, freebsd-bsd-i386,
> et. al.  While this would certainly approach a goal of greater
> universality and kernel/machine independence of the distribution, would
> that gain be worth the effort?

Well, the 'arch' for the NetBSD/i386 is 'netbsd-i386'. Like I said before,
while I'd love to see linux- prefixes to the current architectures, and an
unprefixed arch be supported only as a legacy issue, I don't anticipate it
happening any time soon.

> Perhaps once the Debian/*BSD have stabilized and reach a greater level
> of usability, we can ask these questions again later...

Join us over on debian-bsd@lists.debian.org; we're discussing many of them
now (or, rather, as we run across bits of them that need discussion).
-- 
Joel Baker <[EMAIL PROTECTED]>,''`.
Debian GNU NetBSD/i386 porter: :' :
 `. `'
   `-


pgpJ6hUkf0CWV.pgp
Description: PGP signature


Re: kernel-source == Linux or Hurd or ???

2003-09-24 Thread Ryan Underwood

Hi,

On Mon, Sep 22, 2003 at 08:03:18PM +0200, martin f krafft wrote:
> 
> This is a good point. Debian makes an effort to be kernel
> independent, so why does the kernel-source install Linux?
> 
> I think we should rename to linux-kernel-source, linux-kernel-image
> and so on...

I very much agree with this sentiment.  However, what about system-level
utilities and essential packages?  How does a Debian *BSD system differ from
a Debian GNU system (or for that matter, a Debian GNU/*BSD system), and
how should the dependencies between "bare-metal" packages and the kernel
(whether it be linux-kernel-source, freebsd-kernel-source, etc) be
constructed?

Maybe this is silly, but perhaps the "arch" portion of the apt sources
could also be fine-tuned to include the kernel type.  (I guess similar
to the unique machine strings from config.sub and friends).  A
linux-gnu-i386 distribution, a freebsd-gnu-i386, freebsd-bsd-i386,
et. al.  While this would certainly approach a goal of greater
universality and kernel/machine independence of the distribution, would
that gain be worth the effort?

Perhaps once the Debian/*BSD have stabilized and reach a greater level
of usability, we can ask these questions again later...

-- 
Ryan Underwood, , icq=10317253




Re: kernel-source == Linux or Hurd or ???

2003-09-24 Thread Joel Baker
On Mon, Sep 22, 2003 at 08:03:18PM +0200, martin f krafft wrote:
> also sprach Bernhard R. Link <[EMAIL PROTECTED]> [2003.09.22.1213 +0200]:
> > So your complain reduces in my eyes to an incomplete label.
> > I personally think not having the term "linux" in it more of an
> > issue than having "-debian" in it...
> 
> This is a good point. Debian makes an effort to be kernel
> independent, so why does the kernel-source install Linux?
> 
> I think we should rename to linux-kernel-source, linux-kernel-image
> and so on...

A battle for another day (or year). All I can say is that the only person I
know of who is packaging the NetBSD kernel source (that is to say, 'me') is
using netbsd-kernel-source- (-current for CVS HEAD), in much the
same vein as the current kernel-source- packages.

To date, there are no kernel patches specific to Debian's NetBSD port
(and doing them is a bit of a touchy matter, given the dependance of some
utilities on precise kernel structures; I should probably update the
mini-policy to account for this, at some point).

I do look forward to the day when "not having a prefix/suffice means Linux
only for legacy support reasons", but I don't expect it to happen anytime
soon. :)
-- 
Joel Baker <[EMAIL PROTECTED]>,''`.
Debian GNU NetBSD/i386 porter: :' :
 `. `'
   `-


pgpn3hmJaiKhd.pgp
Description: PGP signature