Re: [sigrok-devel] [PATCH v3 0/2] new config options: averaging and avg_samples

[Martin L. <mar...@ea...>]

On Mon, Jan 26, 2015 at 11:06:16AM +0100, Bartosz Golaszewski wrote:
> 
> What do you think would be the best way to deal with sigrok-cli
> warning about less samples received than expected? Should we remove
> the warning altogether or maybe add a command-line parameter
> '--averages' similar to '--samples', then check it before printing the
> message?

I'm not sure about the averaging behaviour implemented in the demo driver.

In the hardware devices with an averaging feature that I'm familiar
with, enabling averaging doesn't reduce the rate at which the device
displays or transmits data.

Instead, if the averaging setting is set to N, then the data reported
after each measurement is an average of the last N measurements. If that
many measurements haven't been taken yet, the average is taken over as
many as have been.

I think we should stick with that approach, because it means where the
device supports averaging in hardware, the driver can just pass on its
data as normal after applying the averaging configuration.

The user can always drop all but every Nth reported measurement, if they
want the alternative behaviour.

I'm not keen on the "average over all samples" feature either. I don't
know of any devices that would support that in hardware, so we'd have to
implement it separately in every driver, to be consistent with the
behaviour as currently implemented in the demo driver. It seems like
something that would be better placed elsewhere in the architecture.


Martin