Re: ts-adc-debounce: Add interleaved samplings support

Here you are,

patch for final "debouncing" of h2200 devices attached.

Anton, I tried to follow your cosmetic comments in the other mail. Hope I've
got them.

Patch is tested on my h2200 and works well.

Michal, please give it a try also.

Thank you for all your help
Gerhard

On Tuesday 26 June 2007 22:22:26 Paul Sokolovsky wrote:
> Hello Gerhard,
>
> Tuesday, June 26, 2007, 10:23:37 PM, you wrote:
> > Hello Paul,
> >
> > to finish the "debounce" problem for h2200 I'd like to open this
> > discussion again.
> >
> > Independent of the rewrite of ts-adc-debounce we need to
> > set ".delayed_pressure = 1" for h2200 devices and the like in
> > hamcop_base.c. Thats the reason for the first small patch I've attached
> > (hamcop_base.c.diff).
>
> Committed, thanks.
>
> > I assume it should be changed in samcop_base.c too but
> > I can not test it.
>
> Yes, let's leave that to Anton/Milan, but I assume TS on h5000
> works well enough for now, so there's no haste with this.
>
> > Another small patch is for samcop_adc.c. With each reading it writes the
> > measured timeout value into dmesg file. We should avoid this. I've
> > already told you that for h2200 this value is 7 (and sometimes 8). The
> > patch samcop_adc.diff changes the appropriate printk statement to
> > pr_debug.
>
> Ok, I just commented that, it served it's purpose. Anton
> confirmed that it's 0 on samcop.
>
> > On Wednesday 20 June 2007 00:22:13 Paul Sokolovsky wrote:
> >> 1. Let me start with the last querying if we really need any
> >> debouncing with samcop/hamcop. Logs posted by Gerhardshow that in most
> >> cases readings are very stable, and when they are not, that's pen up
> >> cases, which we should have handled by now. Anton, your call.
> >
> > Next thing I've done is to set "num_xy_samples = 1" in hamcop_base.c and
> > "#define Z_READINGS 1" in ts-adc-debounce.c. With a few more
> > modifications in ts-adc-debounce h2200 runs with only 1 reading per
> > sample. I have tested it a few hours without any glitch. The display
> > feels very responsive. Thus my proposal is, to go this way.
> >
> > I've attached patches hamcop_base.c.diff2 and ts-adc-debounce.c.diff for
> > those who want to try it.
> >
> > Sure, going this way would mean we wouldn't do any debouncing at all -
> > but if the result is OK we should go this way.
>
> Yes, I agree. I actually wait for some definitive answer from
> Anton if that will work for h5000 too. If so, we won't need to
> implement interleaved pattern at all for now.
>
> > If you are fine with this I would add num_z_samples variable in
> > tsadc_platform_data make the final modifications in ts-adc-debounce and
> > send you the final patch.
>
> Yes, good plan, please implement that. Handling of
> num_xy_samples = 1 is committed in the meantime.
>
> > Regards
> > Gerhard