Re: The project
On Sun, 18 Jul 1999, Gary Kline wrote:
> So far, just some agreements on a few basics, such as using
> the BSD (FBSD) Linux-``emultation'' rather than mis-invest
> endless months in re-inventing wheels.
To be honest, I think that's a bad idea. The Linux support is (at least in
FreeBSD) rather limited, and even if it worked it wouldn't feel as good
as a "real" system.
But on the other hand, nothing stops us from doing both. Making a "Linux
on BSD"-distribution will probably not take much time at all. Making a
"real" Debian/BSD will take some longer, but most packages compile right
out of the box. What I think will take longest is porting glibc (which is
something that should be done anyway), but any experienced glibc hacker
could probably do it in a weekend.
I think, however, the project will benefit a lot from this, since it will
require some changing of the architecture handling, which the Hurd port
will need to have anyway. (I think Marcus Brinkmann wrote a proposal on
this subject on -private or -devel a while ago)
> I'd like people to begin outlining their own ideas about
> what steps come first, second, third, and Nth-1. What
> are your ideas?
I can give a short brief on what I've been doing so far.
I started with a freshly installed FreeBSD 3.2 system. The standard
distribution of FreeBSD sucks (IMO of course, but since this is a Debian
list I think most people will agree), so I started compiling packages of
my own. The first package was - naturally - dpkg. It took some work
('sysinfo' was missing in the FreeBSD libc), but I had working binaries of
dpkg and the rest up in an hour or so (and most of that time was
Anyway, with a working dpkg that could generate and install packages, I
started doing what else I thought I could need. The following packages
compiled with no or little patching:
...and some more (I have unfortunately no access to my home box from
here). The things we need that I had problems with are:
binutils (the problem was that the make process defaulted to a.out)
gcc (same for this, and is probably easy solved)
The big thing was, of course, glibc. It seems the port that once was done
to 4.4BSD is rather incomplete (almost all of the system specific stuff
seemed to be missing), so it's probably easier to start over from the
beginning. Anyhow, as I said, it's definitely doable. With the BSD libc
sources installed, it's just a matter of writing a bunch of system call
Any volunteer? :)