Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 07 Apr 2009 19:43:26 -0400
From:      Joe Marcus Clarke <marcus@FreeBSD.org>
To:        Tim Kientzle <kientzle@FreeBSD.org>
Cc:        current@FreeBSD.org, Robert Noland <rnoland@FreeBSD.org>
Subject:   Re: Hal and KDM breakage (was Re: KDE4 and input events stalled)
Message-ID:  <1239147806.98664.12.camel@shumai.marcuscom.com>
In-Reply-To: <92cd2ff70904071637h362da63ua13c1f8eca6fc616@mail.gmail.com>
References:  <934e1d760904061455o4736d643o1d07e3292192d94c@mail.gmail.com> <1239075455.1908.36.camel@balrog.2hip.net> <49DACDBD.3030809@freebsd.org> <1239077210.1908.39.camel@balrog.2hip.net> <49DAD429.6090309@freebsd.org> <1239078081.1908.41.camel@balrog.2hip.net> <49DAE987.7090802@freebsd.org> <1239086408.35025.59.camel@shumai.marcuscom.com> <20090407185915.GY31409@albert.catwhisker.org> <49DBA371.3080804@freebsd.org> <92cd2ff70904071637h362da63ua13c1f8eca6fc616@mail.gmail.com>

next in thread | previous in thread | raw e-mail | index | archive | help

--=-Y2kLyzW29+fVSjT2lv6+
Content-Type: text/plain
Content-Transfer-Encoding: quoted-printable

On Tue, 2009-04-07 at 16:37 -0700, Tim Kientzle wrote:
>         See /usr/ports/x11/gdm/files/gdm.in.  This is working for
>         GNOME users.
>        =20
>=20
> I agree that kdm and xdm should have rc.d scripts.  Almost
> every other system is enabled through rc.conf and kdm and
> xdm should be the same.
>=20
> But, there's an enormous amount of documentation
> all over that says to enable kdm or xdm through /etc/ttys.
> I consider this a real POLA problem.  I think that we
> need to continue to support the /etc/ttys mechanism
> for a while longer (and add a comment to the stock
> /etc/ttys to document that xdm/kdm/gdm shouldn't
> be started that way).
>=20
> I also feel pretty strongly that if kdm/xdm/gdm are all started
> through rc.d, then we really have to fix the post-/etc/ttys
> dependency handling.  Your scripts are very confusing and
> basically broken.  For example, "REQUIRE: hald" has absolutely
> no effect.

Not directly, no.  The gdm script does an additional wait to make sure
lshal is working.  So hald needs to be spawned.

>=20
> Of course, I'm still confused about why hald has to start
> after gettys.  I thought the whole point of hald was to monitor
> and publish changes to the system.  Why can't it do the
> same with available consoles?  Is there some missing
> kernel capability that makes this impossible?  Or is
> there an architectural issue with hald that makes this
> harder than it sounds?

It's not a question of what hal is doing, but rather how console-kit
works (and hald depends on console-kit-daemon).  It needs to be able to
monitor each of the active vtys, but won't work if the vtys haven't been
spawned yet.  Bland did some work to correct this in -CURRENT, but it
won't work on other versions.  It was just easier to leave the hack in
place.

If this could be universally fixed in all supported versions of FreeBSD,
then I would be happy to remove the hacks.

Joe

--=20
Joe Marcus Clarke
FreeBSD GNOME Team      ::      gnome@FreeBSD.org
FreeNode / #freebsd-gnome
http://www.FreeBSD.org/gnome

--=-Y2kLyzW29+fVSjT2lv6+
Content-Type: application/pgp-signature; name="signature.asc"
Content-Description: This is a digitally signed message part

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (FreeBSD)

iEYEABECAAYFAknb5RwACgkQb2iPiv4Uz4fbUgCgk9yUf/ltM9BFzBJ1CwDuY5Cr
uZUAn0nSamxuexF0KTY6/gLz//TrL2JS
=nUzM
-----END PGP SIGNATURE-----

--=-Y2kLyzW29+fVSjT2lv6+--




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?1239147806.98664.12.camel>