Re: Patch: detection of Palm with latest HAL no longer working

[Matt Davey <mcdavey mrao cam ac uk>]

Hi Frederic,

I finally got around to integrating your patch.  It is now committed to
SVN, and includes backwards compatibility for the older HAL structures.

I tested it on Ubuntu 8.04, which includes both HAL structures, so not a
100% test, but I'm confident it's working for both cases.

Matt

On Thu, 2008-04-17 at 12:16 +0200, Frederic Crozat wrote:
> Le mardi 25 mars 2008 à 10:24 +0000, Matt Davey a écrit :
> > Hi Frederic,
> 
> First, I apologize for my slow response, I got busy in finishing
> Mandriva 2008 Spring and I also attended conference in the mean time.
> 
> > 
> > On Fri, 2008-03-21 at 19:10 +0100, Frederic Crozat wrote:
> > > Hi,
> > > 
> > > we switched in Mandriva to hal 0.5.11 (rc2 at the moment, final is
> > > pretty imminent) and info.bus is no longer advertised as part of usb
> > > devices (key has been deprecated).
> > 
> > Thanks for pointing this out.  I would like to support hal backwards
> > compatibility for the moment, so how about we check for pda.platform
> > first, and then fall back to devices.xml and info.bus (or another,
> > supported, way to query USB vendor/device IDs)?  That way we avoid
> > requiring the most up to date HAL version for gnome-pilot - if someone
> > wants the new bluetooth support on an existing installation, they
> > shouldn't need to upgrade HAL.
> 
> Yes, fallback is probably better than my dirty search and replace
> patch ;)
> 
> > > The attached patch changes this check with a check on pda.platform.
> > > 
> > > Those keys have been recently added in hal-info to detect all known
> > > Palms (supported by pilot-link) so device permissions can now be handled
> > > by hal / ConsoleKit / PolicyKit.
> > > 
> > > I think a good patch or hal_device_added logic (including usage of
> > > devices.xml) could be removed since most of the information needed for
> > > device detection is now directly available from HAL, including the right
> > > device to use (when using visor module).
> > 
> > How robust do you think the usbserial tty device identification can be?
> > I'd love to be proved wrong, but I'd be surprised if this had actually
> > been solved!  There have been many attempts over the years, and if you
> > have a device that behaves differently and confuses HAL, you'd be pretty
> > annoyed if there isn't a workaround.
> 
> Well, ATM, hal doesn't do any USB tty device identification. It only
> uses USB ids. And we know usb ID are not reliables for USB tty, since
> several Palm devices share same usb ID and uses different tty.
> 
> My suggestion was for people using libusb pilot-link and not for visor
> module based pilot-link.
> 
> > I do agree that letting HAL do more work is the right thing to do, but
> > it has to make life easier for the users.
> > 
> > A would be useful to rework the config applet to do more automated
> > detection of usb/serial/net/bluetooth devices - listen to everything and
> > configure the first palm that connects.
> 
> Then it would need to be integrated with nautilus new "volume manager"
> so it would be auto-started when a palm is started but was never
> configured.

Matt Davey        "I do not agree with a word you say, but I will defend	
mcdavey mrao cam ac uk      to the death your right to say it".  Voltaire