[PATCH 2/2] Correctly calculate SAC rate in the presense of surface events

[Linus Torvalds]

On Sun, Feb 24, 2013 at 11:10 AM, Dirk Hohndel <dirk at hohndel.org> wrote:
> Linus Torvalds <torvalds at linux-foundation.org> writes:
>> What do people think we should show for the dive duration field in the
>> dive list for that last sac-test dive? We spent 30 minutes on the dive
>> end-to-end, but only 25 minutes actually under water. Is it a 30-minute
>> dive, or a 25-minute one? Is the average depth 10m (which is what we use
>> for SAC calculations) or 8.33m (which is the average over the whole 30
>> minutes)?
>
> I think it's a 25 minute dive with an average depth of 10m.

Ok, that last patch I sent makes it so.

As to the whole discussion about exactly how to estimate SAC rate -
there's nothing we can ever do that is perfect. We can't detect
temperature-induced pressure changes ("is the cylinder cooling down,
or is the diver breathing"), and while it's true we can detect huge
spikes (sensor problems, possible free-flow), I'd hate to try to be
clever about that, especially since we wouldn't see things like that
anyway if the pressure data was given by hand for just beginning end.

So the best we can do is take the amount of gas used and turn it into
a SAC rate given the depths we see. And from everything I've ever
seen, for extended surface times, people do tend to breathe the air
around them. And I think the code I wrote is the best we can do.

Now, one thing I would personally like to see is the ability to see
SAC rates not for the whole dive, but for a certain *part* of the
dive. We don't have that kind of interface, though. And it really is
all about the interface, if we had some way to select a range, I'd
happily write the code to calculate SAC-rate during that range.

               Linus