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

Re: debinst: native hurd installer

On Mon, Sep 18, 2000 at 09:33:51AM +1100, Glenn McGrath wrote:
> Ive been playing with the Hurd, and talking to a few people (notably
> Marcus Brinkmann whose cc'ed), i have a few comments id like to make
> here.

Thanks for summarizing this, Glenn. I will make some small corrections and
further comments.
> The Hurd doesnt use linux kernel style modules at all.

It is more likely for us to use user-space drivers, but for performance
reasons all direct hardware support is in the microkernel. Also, nobody is
really interested in making linux style modules work.

> The Hurd has no use for hardware detection, it either works or it
> doesnt, but i guess the hardware present
>  will determine wether translators need to be used, and what
> configuration they will need.

Everything is still very simple for us ;)

BTW, graphic cards are a different story of course, but I think that it will
be very difficult to implement hardware detection with GNU Mach. Anyway,
hardware detection is probably unportable. Anybody thought about a hardware
detection outside of the OS, as a stand alone program?

> Currently the easiest way to install the hurd is to unpack a ~20MB (i
> think)

It's 14 MB, but only because it contains the full perl (because ENOENT
isn't in perlbase). In other respects this tar file is very similar to the
base tar file in the Linux install.

> tarball and then a general purpose configuration script is ran
> that configures... stuff.... (im in the dark a bit here).

It sets up translators (incl. device files) and configures the Debian
packages, which are only --unpack'ed in the tar file (for cross installation

> There is the cross-install method of installing it as well which i
> havent tried.

cross-install basically creates the unpacked tar file. 
> I get the feeling that any help is very much appreciated, if we came up
> with a way to install hurd from linux then they would indeed be happy,
> but they wouldnt consider it to be a finished product, they would feel
> the need to hack at it untill it is a native Hurd installer. (maybe im
> being too judgmental here)

Well, a mini linux with an installer would certainly be a big improvement
over the current situation and gladly accepted. When I say that in the long
term we want to be self contained, then I talk about how we want it to look
in one to two years.

> I dont think making a native hurd installer would be much more
> difficult, i started to have a go at converting busybox, and i think
> much of it will work without much effort. Slang, ae, wget and dpkg (of
> course) are already supported, thats about all thats on the root disk as
> i remember.

There are probably some bugs in the Hurd to flesh out, and you won't be
able to fit everything on one disk. You need two (one boot disk, one with
the root fs) at least. The basic Hurd is a bit fatter. I had some efforts
going in this direction, and am certainly willing to pick them up again.

Thanks a lot,

`Rhubarb is no Egyptian god.' Debian http://www.debian.org Check Key server 
Marcus Brinkmann              GNU    http://www.gnu.org    for public PGP Key 
Marcus.Brinkmann@ruhr-uni-bochum.de,     marcus@gnu.org    PGP Key ID 36E7CD09
http://homepage.ruhr-uni-bochum.de/Marcus.Brinkmann/       brinkmd@debian.org

Reply to: