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

Base set is coming closer...



Hello,

this is my daily progress report. If you are not interested in whiny cross
compilation stories, stop reading now :)

1) Beside the GNUmach package, there are now the following packages up and
   running:
glibc2 -> glibc2, glibc-dev, glibc2-doc  (Version 2.0.4)
hurd -> hurd, hurd-dev                   (Version 0.2 = 19970613)

Both are the same sources as the current gnu-0.2 binary dist or the hacker
kit, but recompiled and packaged for Debian. The packaging rules could
really need an improvement, but for now it sort-of-works.

This is really all that is needed for the Hurd to get a functional system.
Well, people who like more convenient ways to start programs than adding
them to /boot/servers.boot script would enjoy a shell, too. And some
programs (unimportant ones like "rm", "ls" and "cp") are also useful
sometimes. But those are the packages that are hurd-only at the moment.

(Don't run the Debian package checker "lintian" on them. Doing so would
reveal a lot of Debian policy violations :)

I would suggest that anybody who is doing cross compiles should focus on the
"essential" packages and the base packages (this would be Santiago and me
for now, I hope other will join us as soon as we have the crucial base
ready).

This is my personal urgency list:
bash,
fileutils,
gzip (Santiago did it already),
sed (Santiago did),
grep (Santiago again!),
tar,
perl-base,
sysvinit (? is there some hurd-ism to care about?),
bsdutils (script!! for logging --- we can't scroll up the terminal),
textutils (guess it --> yep, Santiago),
ncurses,
base-passwd (Santiago did),
less ("more" is part of util-linux),
update (? will it run on the Hurd? Functionality important else data loss), 
dpkg & co. (I did, have to test it and fix architecture),
base-files (pretty easy, maybe we need Hurd specific extensions though),
findutils,
shellutils, 
hostname (done by Santiago),
diff,
not to speak of the libraries and other stuff they depend on
etc
    etc
        etc

(This should be sufficient to keep us occupied for a few weeks or months, I think).

The following are probably not needed on the hurd:
ldso (is in "glibc2"), mount (use translators instead), login (is in "hurd").

2) My packages really need a clean up. If you ask me, the Makefiles
for those sources need a rewrite, too (they work, but...).

However, the following packaging issues can be discussed:

a) What can be stripped, what not? As I see it, only ld-2.0.4 may not be
stripped, *.so files can be strip-unneeded'd, *.a files can be strip-debug'd.

b) What can happen during an upgrade? Gordon, I need your help here!
Debian's upgrade mechanism supports several scripts, that are run before
removing, after removing, before installation and after installation. You
can query what is happening, a new install or an upgrade. After unpacking
all packages to be installed, they are configured, which is a seperate step.

So, you can stop/restart daemons etc. How robust are translators during such
an upgrade/replacement?

What happens when a translator is connected to a file, and the file is
contained in a package (for example /servers/exec). Is it possible to
include an empty file in the package, or should it be created in the post
install script? This issue is not urgent, but needs to be adressed at some
point in the future.

3) If somebody else than me is also taking the route to build a hurd system
from scratch (or using my packages), be sure that you have empty files
"/servers/exec" and a directory "/servers/socket", or the booting will fail
horrible (and the error message is not really helping). Report to bug-hurd
is under way.

==========================================
You also need a link "ld.so" -> "ld.so.1", or init will not find the linker.
==========================================
Doh! This will be certainly fixed in my next glibc2 upload.

4) My plans are: porting bash (somewhat important!), then fileutils (so I
can replace "echo *" with "ls" again !). This is also important because the
MAKEDEV script uses chown etc to operate.

Then I think I'll try to understand why the current hurd snapshot does not
work. A fresh snapshot from the developers would be fine, too, if they think
they have fixed some things (Gordon, could you please check this?).
Otherwise I'll just do incremental patches I think. I'm not sure how this
will work. Maybe it will not work at all, then we have to wait for a working
developer snapshot (what about CVS, guys? Or would anyone else be able to
provide an arla gateway until Michael is online again?).

Thank you for reading so far,
Marcus

-- 
"Rhubarb is no Egyptian god."        Debian GNU/Linux        finger brinkmd@ 
Marcus Brinkmann                   http://www.debian.org    master.debian.org
Marcus.Brinkmann@ruhr-uni-bochum.de                        for public  PGP Key
http://homepage.ruhr-uni-bochum.de/Marcus.Brinkmann/       PGP Key ID 36E7CD09


-- 
To UNSUBSCRIBE, email to debian-hurd-request@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org


Reply to: