Meaning of GF settings

[Dirk Hohndel]

Linus Torvalds <torvalds at linux-foundation.org> writes:

> On Tue, Jan 8, 2013 at 6:37 AM, Robert C. Helling <helling at lmu.de> wrote:
>>
>> ok, here we go. This does look somewhat reasonable to me. Please check that
>> it gives sensible results for your test cases!
>
> I have a test-case that this does not give sensible results for. I
> don't think it's your deco code, and I'm fairly certain it existed
> before, but I felt that maybe both you and Dirk care.

Certainly.

> The test-case is having two events on the same second. I don't know
> quite how that happened, but I seem to have profiles like that
> especially from my old Gekko days. I *suspect* it might be due to a
> temperature change event, causing the gekko (or libdivecomputer) to
> have given me two different samples. Or it could be just an effect of
> me merging dive information from several sources (ie the suunto SDE
> file along with possibly old libdivecomputer imports etc)

My guess is that this is from a bad merge back when you merged samples
instead of doing different divecomputer files. But either way, we need
to handle stuff like this.

> Anyway, I'm attaching one such dive for your viewing. It's my Blue
> Hole (Belize) dive from a few years ago. When there are these kinds of
> events, there's a discontinuity in the computed deco graph where it
> goes to zero for that event, and then comes back again immediately.
> See this happen at the 12 minute mark, and then again at 19 minutes.
>
> Dirk - any ideas? Some "time delta is zero, so create zero deco data"
> special case?

I'm not aware of that, but I'll take a look. My best guess is that we
should simply skip one of them for deco calculation :-)

/D