[Top][All Lists]

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

Re: A GNU Distribution

From: Daniel Martin
Subject: Re: A GNU Distribution
Date: Sat, 31 Mar 2012 14:40:33 +0100

(There are people far more qualified than me who can answer this but I'm
not sure they are active at the moment. I've provided this response but
please be aware that I defer to the greater knowledge of others on this
list in the event they are able to offer a better reply.)

On Fri, 2012-03-30 at 19:19 -0700, Jason Self wrote:
> Now that Linux-libre has been dubbed an official GNU package it seems to me 
> that the GNU Operating System now has a fully functional kernel. Perhaps this 
> means that there could now be an official release of the GNU Operating System?

Sort of. There's already been quite a bit of work on the GNU system, as
originally envisaged, running on the Hurd. The Hurd replaces a unix
kernel but is not identical to one. There are nice pretty things that in
theory can be done using the Hurd that can't be done on a traditional
monolithic kernel. The GNU system to the best of my knowledge had plans
to make use of these possibilities throughout it's design. (For example
in its package manager.)

The advent of FUSE in Linux might make some of these things possible on
Linux though. (???)

> I'm starting this thread to open a discussion of what sort of shape such a
> distribution might take and what needs to improve within GNU in order to do 
> it.
> My idea is that it would be source-based where people compile the operating
> system from scratch, which brings up my first idea...

What's the benefit of making people compile their own packages?

I've always thought it would be nice to install source-code alongside
binaries so that users could _easily_ exercise their right to modify
their software. (Without breaking out of the package management system.)
If such a feature belongs anywhere it belongs on the GNU system. I don't
think anyone else shares my view on this though.

> For starters it seems to me that there needs to be some way to install the
> operating system onto bare metal, unless people are to be expected to 
> bootstrap
> it using an existing system?

So far, you do indeed need an existing system to install. I don't think
a system installer is the most important thing missing from the GNU
system. Anyway, if memory serves Debian has an installer for its
GNU/Hurd distro. The GNU system could make use of that, at least to
begin with.

> In addition, people could choose whether they want to use Hurd or Linux-libre
> during the compile process. Another option might be that both kernels could be
> installed at the same time and people could select between them on boot using
> the GRUB menu. It seems that this type of setup would add additional 
> complexity
> but that it would also not be impossible.

I _suspect_ that this would be very complicated and make it difficult to
keep the distro up-to-date as I think you'd be limited to the lowest
common denominator or the available kernels. If you're going to put
yourself through that you may as well let people run the system on
FreeBSD's kernel as well.

I think the only reason to do this would be for political reasons; to
show that the system can be run on a variety of kernels besides just

> Thoughts? Ideas? What else needs to be improved within GNU to have our own
> distribution?

I don't think that GNU is in the best position to produce a GNU/Linux
distro. Would the system offer anything that isn't already available
from other distros?  Again, I think the only reasons to do so are
political. The only sane way to go about doing this would be to fork an
existing distro and I think that would be viewed quite sceptically by
the community (and rightly so).

There's a real lack of man-power for this project and I think that's
only going to change if we are going to build something new and
exciting. Rather than yet another GNU/Linux distro.

By way of example, the ngHurd (or whatever it's called now) project
always seems to me to get more attention from new hackers that the
original Hurd project simply because it seems new and exciting.

Anyway, there's my opinion for what little it's worth.

This list has been very quiet for a long time but hopefully someone will
chime in and offer better info / advice than I can.

Best regards,


PS: You realise that there are GNU system (pre-?)releases that you can
have a play with, right?

reply via email to

[Prev in Thread] Current Thread [Next in Thread]