Re: Free software telemetry and the Guix System

[Bengt Richter]

Hi all,

On +2021-05-14 16:52:25 -0400, Leo Famulari wrote:
> On Fri, May 14, 2021 at 06:55:34PM +0000, Cook, Malcolm wrote:
> > > If these claims are true, then I think this is quite satisfactory for
> > > our purposes. I wouldn't even object to enabling the telemetry code via
> > > the CMake build-time option, as long as it's "opt-in", i.e. that each
> > > user must explicitly enable it, and only after being made aware of the
> > > consequences of doing so.
> > > 
> > > What do you think?
> > 
> > My 2 cents:  I think the Audacity model is exemplary and your 
> > interpretation is spot on.  I personally want the option of enabling such 
> > telemetry, as it may well serve my needs and may also give the developer 
> > valuable usage and/or crash info which is the least I can provide in return 
> > for such a great FOSS app as Audacity.
> 
> +1
> 

My 2 cents:  :)

I like options, but I would feel more secure if it were implemented in a 
separate,
dynamically linked when opted-in,
    some-implementation.so
which I could get the kernel to prevent access to, e.g. by
    # chmod 400 some-implementation.so

-- 
Regards,
Bengt Richter