[Tinyos Core WG] Unit of time in TOS2

[David Gay]

On 10/20/06, Cory Sharp <cory at moteiv.com> wrote:
> On 10/20/06, David Gay <dgay42 at gmail.com> wrote:
> >
> > One comment: the introduction has a paragraph about accuracy which
> > explicitly says that the timers may not be super accurate (i.e., that
> > 1/1024s might well be 1/1000s anyway, the 1/32768s is definitely
> > something else on mica family platforms). So worrying too much about
> > the names doesn't seem that useful to me...
> >
> > Which does possibly bring up the question of whether we want to
> > mandate better accuracy for TMilli... (but probably also a whole can
> > of worms about drift, crystal accuracy, etc, that we may not care
> > about...)
> >
> > David
> >
>
> Ah! Mandating accuracy for TMilli reminds me of our original plan (?) from a
> year and a half ago.
>
> 1) Leave TMilli and TMicro as decimal names
> 2) Platform implementations may very well be off by ones of percent
>  3) Accuracy is not guaranteed by the TEP, anyway, because
> 3.a) Variations in hardware components between platforms
> 3.b) Clocks the internal clocks may significantly drift with
> temperature/voltage, dwarfing the 2-5% binary/decimal variance
>
> This brings us back to the network time and timers TEP -- if you really want
> to say "millisecond" and have it mean something reasonably accurate, you
> need to be synchronized with the network.  Those components and TEP do not
> yet exist, but it's the right way to solve the problem -- building a timer
> system built on synchronized timed.
>
> The implication for TEP 102 is then we need to actually soften up the
> language about "binary" units (though perhaps leave some discussion in there
> about it), not strengthen in by renaming the precision tags.

That all sounds good to me.

David