[ltp] Re: hdapsd during Intercity Express ride

Micha Feigin linux-thinkpad@linux-thinkpad.org
Fri, 16 Oct 2009 23:35:56 +0200


On Fri, 16 Oct 2009 01:14:34 +0200
Jan Girlich <vollkorn@cryptobitch.de> wrote:

> Am Donnerstag, den 15.10.2009, 15:00 +0200 schrieb Elias Oltmanns:
> > Jan Girlich <vollkorn@cryptobitch.de> wrote:
> > > Am Montag, den 12.10.2009, 20:48 +0200 schrieb Martin Steigerwald:
> > >> I am currently traveling with a Intercity Express - mail will be sent 
> > >> later, cause no internet access in the moment - and found a cumbersome 
> > >> problem: There have been extreme lags during most common operations.
> > >
> > > I had the same problem during a bus ride. The system is rendered useless
> > > because of vibrations like from car engines.
> > 
> > From a mathematical point of view, I have serious doubts that the logic
> > implemented in hdapsd is the right thing(TM). The problem is, as far as
> > I'm aware, that we don't know exactly what the data returned by the
> > accelerometer is supposed to represent, i.e. whether any kind of
> > mathematical preprocessing is performed. I seem to remember that Shem
> > performed some experiments when (s)he made changes to hdapsd. Whether
> > real rides by bus or train had been among those tests, I don't recall.
> > 
> > >> On Windows there is an option to ignore continued vibrations from such a 
> > >> situation - we checked with a laptop from a co-worker. Is there such a 
> > >> knob for hdapsd? I thought about
> > >
> > > I don't know about any options, but I haven't looked into it, yet.
> > >
> > > But I also think the thresholds are set way too sensitive and hdaps kind
> > > of 'overreacts'.
> > 
> > What thresholds exactly? Have you tried the -s option yet? If you work
> > on bus / train frequently, it might be worthwhile fiddling with this
> > option and reporting your results to this list.
> 
> I did not have a look into hdapsd options at the time of writing the
> former email. I just did so and will try using the -a option on the same
> bus ride which rendered my system useless tomorrow and report here.
> 
> When I said 'thresholds' I meant the sensitivity level.
>

Personally I had to play with the thresholds to get to the point where it reacted to turning over (simulated fall) and large bumps but with the defaults the train ride rendered the machine useless otherwise, even with the -a option
 
> Cheers
> Jan