/tmp/hald-local and /tmp/hald-runner

[Joe Marcus Clarke]

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

David Zeuthen wrote:
> On Tue, 2006-05-02 at 10:30 -0700, Artem Kachitchkine wrote:
>> Joe Marcus Clarke wrote:
>>> These two directories are vital for HAL to startup.  Yet I do not see
>>> them created anywhere.  If they are not created manually prior to
>>> starting hald, then hald will die.  I checked the dbus code, but they do
>>> not appear to be created there, either.  I'm wondering how this works on
>>> Linux.  My thoughts on fixing this were to add mkdir(2) calls to
>>> hald_dbus.c and hald_runner.c just before dbus_server_listen() is
>>> called.
>> Aren't they abstract sockets in Linux? Abstract sockets don't have 
>> filesystem representation, so they are not affected by the problem you 
>> described. 
> 
> That is correct.

Yes, FreeBSD doesn't support abstract sockets.

> 
>> On Solaris, I create these directories in the hal startup 
>> script. 
> 
> Maybe hald should create them when starting up?

That's what I was going to do since we're still in the phase where we
recommend users run hald with --daemon=no --verbose=yes.

> 
>> I was also thinking about moving them somewhere safer, such as 
>> /var/run/hal.
> 
> Sounds good with me.

Okay, so adding mkdir(2) calls to hald_dbus.c and hald_runner.c is the
way to go?  If so, I can submit a patch that changes the socket location
and creates these directories if you want.

Joe

- --
Joe Marcus Clarke
FreeBSD GNOME Team	::	gnome at FreeBSD.org
FreeNode / #freebsd-gnome
http://www.FreeBSD.org/gnome
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2 (Darwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFEV57hb2iPiv4Uz4cRAqK0AJ9+W86QQ4aZNctraUgge1nVQ5flJACeMz3m
dgNhgr01aZGl+jJZf4oXOvc=
=vqLx
-----END PGP SIGNATURE-----