Re: possible odd acceleration scaling

[Bastien Nocera <hadess@hadess.net>]

On Wed, 2016-09-14 at 08:28 -0700, Srinivas Pandruvada wrote:
> On Wed, 2016-09-14 at 19:20 +0530, Ritesh Raj Sarraf wrote:
> Hello Srinivas,
> 
> On Wed, 2016-09-14 at 14:58 +0530, Ritesh Raj Sarraf wrote:
> > 
> >  
> > And linked are debug logs with both, 1.1 and 1.2
> > 
> > https://people.debian.org/~rrs/tmp/1.1.log
> > https://people.debian.org/~rrs/tmp/1.2.log
> > 
> 
> 
> After reverting your changes[1], and applying Bastien's change [2],
> it is better
> as in now it can detect all orientation types. But they do not align
> with the
> actual physical orientation of the device. They are flipped to the
> actual
> orientation of the device. Logs linked [3]. Tested on Linux 4.8-rc6
> 
> rrs@learner:/var/tmp/Debian-Build/Result$ monitor-sensor 
>     Waiting for iio-sensor-proxy to appear
> +++ iio-sensor-proxy appeared
> === Has accelerometer (orientation: right-up)
> === Has ambient light sensor (value: 0.000000, unit: lux)
>     Accelerometer orientation changed: normal
>     Accelerometer orientation changed: right-up
>     Accelerometer orientation changed: bottom-up
>     Accelerometer orientation changed: right-up
> 
> 
> I was surprised why the sorting function didn't work (it should be
> qsort compliant). We need some way not to depend on the order of
> scan_elements read from file system. So we need to fix this.

My guess:
-       return (int) (info_1->index - info_2->index);
+       return ((int) info_1->index - (int) info_2->index);

Ritesh, could you try to change those lines in compare_channel_index()
in src/iio-buffer-utils.c and capture another log if it fails?

Cheers