[Tinyos Core WG] Meeting: 10/18
Cory Sharp
cory at moteiv.com
Wed Oct 18 14:44:46 PDT 2006
On 10/18/06, Philip Levis <pal at cs.stanford.edu> wrote:
>
> On Oct 18, 2006, at 1:42 PM, Martin Leopold wrote:
>
> > Hi Phil.
> > I missed the first meeting - not a god way to get started. I can se
> > you
> > are in crunch mode to get 2.x released and I talked with Cory about
> > some
> > of my concerns for the timers - he asked me to bring it up and I was
> > intending to do so, but alas I dint realize the meeting was so
> > early in
> > the morning =[.
> >
> > Anyway - what I wanted to bring up was the name of the timer precision
> > TMili gives 1/1024 s not 1/1000. This is a bit unclear from the TEP
> > and
> > the name is sort of misleading.
> >
> > So before I start a heated debate while tep is _just_ about to
> > release:
> > do you think there is any humanly possible way that the core group
> > would
> > come to a consensus to change the name before 2.x releases?
>
> I think there were some email exchanges about this.
>
> I'm personally of the opinion that TMilli should stay. We should also
> have a T1kHz or something for the true millisecond case. That being
> said, TEP 102 should definitely and clearly state that TMilli means
> 1/1024 and not 1/1000. Milli is what's been used in TinyOS since the
> beginning and it's been OK. There are 1-2 questions a year on -help,
> but most of the time it's fine.
>
> What would the name for 1024Hz be? T1024Hz? That seems really
> unwieldy for the common case...
>
> Cory, David?
>
> Phil
>
I agree that the issue comes up infrequently. Especially since the TEP
asserts at one point that the values are in "binary", but later on that they
are also only approximations. Even so, there is something to be said for
not conflicting with an international standard (redefining milli from 10^-3
to 2^-10) when a suitable international standard may already exist.
The standard SI prefixes do actually include the concept of positive powers
of 2 in addition to powers of 10, as described here
http://physics.nist.gov/cuu/Units/binary.html
Found from here http://en.wikipedia.org/wiki/Si_prefixes and
http://en.wikipedia.org/wiki/Binary_prefix
.
Even though the standard does not apparently cover negative exponents, a
pattern that we can apply is discernable: in the factor name, substitute the
second vowel and its preceeding consonant with "bi" (prononced "bee"), so
"kilo" becomes "kibi", mega becomes "mebi", giga becomes "giba", and so on.
Given that pattern, we could rename "milli" to "milbi" and "micro" to
"micbi" for precision tag names TMilbi and TMicbi.
We do have an existing convention, and most people don't seem to be
concerned, but if there is an alternative to be considered, I'd offer up
"TMilbi" in place of "TMilli".
Cory
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.millennium.berkeley.edu/pipermail/tinyos-2.0wg/attachments/20061018/fa12bad7/attachment.html
More information about the Tinyos-2.0wg
mailing list