Re: ipw srcipts for hal

[Dan Williams <dcbw redhat com>]

On Wed, 2007-06-27 at 14:15 +0100, Bastien Nocera wrote:
> On Mon, 2007-06-25 at 12:13 -0400, Dan Williams wrote:
> > On Mon, 2007-06-25 at 16:09 +0200, dragoran wrote:
> > > 
> > > 
> > > On 6/25/07, yelo_3 <yelo_3 yahoo it> wrote:
> > >         > ok, but this does not solve the problem of multiple
> > >         killswitches (will
> > >         > show up with multiple cards) because both will have
> > >         > /org/freedesktop/Hal/devices/ipw_wlan_switch as uid.
> > >         Yes the previous shell script didn't solve the problem...
> > >         sorry. This might mean that the UDI should contain the
> > >         interface name as you were saying 
> > >         The C code misses the setrfkill section, and a !=null check
> > >         when you fopen the file.
> > > 
> > > ok here is a new version.
> > > it implements setrfkill too and uses g_strdup_printf instead of
> > > sprintf.
> > 
> > I don't think argc == 3 is valid for the setrfkill check, since the
> > number of args will still be 2...  I'd just put a check before the
> > libhal_ctx_init() that does:
> > 
> > if (argc != 2) {
> >     fprintf (stderr, "Usage: ipwWirelessCtl [getrfkill] [setrfkill [1|0]]\n");
> >     return -1;
> > }
> > 
> > or something like that, and get rid of the argc checks for getrfkill and
> > setrfkill.
> > 
> > I think we should actually just reparent the device to be a child of
> > Computer.  I also think the script should just rfkill _everything_, and
> > that it should return '1' if _any_ ipw radios are off.  This script is
> > really only a stopgap until the _real_ kernel rfkill interfaces are
> > complete, and then most these problems go away.  So instead of trying to
> > overengineer the whole thing, I think it should work like this:
> > 
> > a) .fdi file adds _one_ rfkill device if any ipw cards are found
> > b) 'ipwWirelessCtl getrfkill' checks all ipw devices for rfkill status,
> > and if one of them is killed, it returns 1
> > c) 'ipwWirelessCtl setrfkill 1' kills _all_ ipw devices, while
> > 'ipwWirelessCtl 0' re-enables _all_ ipw devices
> > 
> > Sound OK?  That way we also don't have to figure out how to unique-ify
> > the device name, which the future kernel patches will handle for us.
> > 
> > I've attached an updated .fdi file for ipw devices that excludes Dells,
> > and makes only _one_ killswitch device.
> 
> Please attach the killswitch to the device itself, not the computer.

I changed it to be on the computer object, because this is all really
just a hack until the kernel rfkill patches land and we get _signals_,
which is what I really really want.

I don't really care either way.

Dan