USB CDC ACM modems proliferation in 10-modem.fdi - please stop

[Dan Williams]

On Fri, 2008-11-07 at 18:43 +0200, Alex Kanavin wrote:
> 2008/11/7 Danny Kukawka <danny.kukawka at web.de>:
> > IIRC there were some objections against probing tty devices via HAL in the
> > thread where this was already discussed. What about these objections?
> >
> > Because of this it's IMO nothing for the upcomming release because it needs
> > more testing before put it into a release.
> 
> As far as I understood, Dan Williams was happy with the prober as it
> is, and was about to add it to hal. I'm not sure why things got stuck
> at that point.

The objection was that some modems fail probing when they have a PIN
code set.  Turns out all of the modems that behave this way that I've
seen respond to ATI with the necessary information, so we will just not
care about any that don't.

As noted above, the prober will be going into a udev-extras package, and
will be udev addition with some udev rules.  The prober will populate
the udev database, and HAL will in turn get a callout to read the udev
database and transform the properties there to HAL device properties.

This is a forward path to enable both DeviceKit-style things that are
more closely based on udev, and still allow the prober to work with HAL.

Furthermore, it allows parallel device probing with earlier versions of
HAL, because udev probes in parallel, but HAL currently does not and may
never do so.  Thus, if you have a modem that exposes 4, 5, or more ports
(yes, they exist), with the udev prober, you'll see only a 4 second
startup hit, but with HAL you'll see a (4s * #ports) hit, which just
sucks.

Dan