Re: [GNUnet-developers] experimental debian packages

[Christian Grothoff]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Saturday 08 February 2003 04:10 am, Glenn McGrath wrote:
> I have some experimental debian packages available at
> http://people.debian.org/~bug1/gnunet
>
> You will find see debs of libextractor and libgcrypt as the GNUnet debs
> depend on them.

Great. I'm now running Debian on one of my machines, so I'll even be able to 
test them :-)

> They still need a bit of work, currnetly GNUnet runs from init.d as root
> which i see is discouraged, i will change it when i decide what user it
> should be. Its enabled by default as well, i will put something in to
> make it a more conscious decision to start so users are more aware its
> going.

Well, the "-u" option needs some work (see Mantis) anyway...

> Its setup to use system wide config, i think people will get sick of
> type -c /etc/gnunet.conf, but debian discourages user specific config
> files if there is an alternative.

I think we should probably change the GNUnet code to fallback to 
/etc/gnunet.conf if ~/.gnunet/gnunet.conf is not found. Is /etc/gnunet.conf a 
good location? Or should it be /etc/gnunet/gnunet.conf? Or 
/var/gnunet/etc/gnunet.conf? Where should the "shared", systemwide gnunetd 
databases be placed anyway? I would think /var/gnunet/ would be the most 
appropriate thing. Any other opinions?

> I think it would be advantagous to have it running system wide anyway as
> all users can pool the archive, what are the technical disadvantages ?

Well, if users *index* files and then move or delete them, they "pollute" a 
system wide database. The index & move is a problem even without that, 
though. The best solution I can come up with would probably be to have 
gnunet-insert *move* the files to /var/gnunet/shared and make the directory 
owned by "gnunet:gnunet" and 775 permissions, make gnunet-insert SGID or 
SUID, add possibly using quotas. This could prevent users from indexing files 
and then remove them from sight. Other than that, I can see no disadvantage 
from users of the same machine sharing a gnunetd process.

> So if you try these packages keep in mind that nobody else has tested
> them, and feedback is most welcome.
>
> They wont become offical debs until after the next libgcrypt release and
> theyve been tested a lot more.

Definitely a good idea :-)

> And Christian, I see now the changes i need to make to the config file
> handling, namely variable substitution.
>
> I think having some stats in gnunet-gtk would pretty nice as well.

Well, either do it yourself or add a feature request to Mantis to make someone 
else do it, I guess. I think Igor's priority for GTK will rather be segfaults 
& memory leaks for now, but I could be wrong...

Christian
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: For info see http://www.gnupg.org

iD8DBQE+RcE/9tNtMeXQLkIRAsUFAKCas6ygZOzec3YAUl1+cH+FzMt9nwCfXbdU
Yieximb9W6sOh+tH1qaZBOE=
=rHcC
-----END PGP SIGNATURE-----