[Tinyos-devel] Request for comments on TEP 103, due August 10th

[Joe Polastre]

> Unfortunately, that leaves a typical developer in a bit of a bind: if
> the only choices are HAL (optimal performance) or HIL (optimal
> independence), then there is no middle ground with tradeoffs. The
> paper -- and TEP 2 -- certainly don't take so strong a stance. In
> fact, the HAA paper proposes Log, Config, and Block! E.g.:

The point of my email was as follows: The HIL is allowed to sacrifice
performance in the interest of independence.  Independence is achieved
through "typical hardware services required in a WSN application."

If David's HIL proposal leads to more interfaces above a common
"narrow waist" than the current Conflig/Log/Block, then I firmly
believe it is the correct choice for an HIL layer and it provides the
interface for typical hardware services required for a WSN
application.

> The HAA paper proposed them....

Yes, they also proposed the wrong Timer and ADC interfaces too, which
have changed with more experience.  I think this is a clear case where
additional experience can lead to a better design.

As of the writing of the paper, no one had been able to get a file
system working in a cross-platform implementation.  David Moss, afaik,
is the only such individual to achieve this feat.

-Joe