[PATCH 4/6] Add support for libdivecomputer DC_FIELD_DIVEMODE

[Dirk Hohndel]

On Sun, Nov 09, 2014 at 11:11:44AM +0100, Jef Driesen wrote:
> >OK, I ended up just implementing this… it’s pushed to master. In a broken commit
> >that I pushed earlier and then a fix once I realized that this didn’t compile
> >with older versions of libdivecomputer - but since it had been in the public
> >tree for a while it seemed wrong to force push a fixed commit, so my stupidity
> >will stay visible for the world…
> >
> >Please test.
> 
> I did a quick test using a Cobalt 2, and I think there are some issues.
> (Note that there was a bug in libdivecomputer too, for which I just pushed a
> fix.)

That obviously isn't included in the Windows binaries in case anyone is
using those.

> I attached the raw data for one such dive. It has 5 tanks configured. Four
> metric and one imperial (in CUFT at BAR mode). Only one of the metric tanks is
> used. See the libdivecomputer xml output below. When downloading this dive
> into subsurface, this looks wrong:
> 
>  * None of the tanks has volume or working pressure.

That sucks, so my code is wrong. That's always a pain to debug when you
don't have access to the hardware

>  * The O2 percentage for air is missing. I'm not sure if this is by design
> or not, but I expected to see 21% here.

That's by design. Air is special. No gasmix info means you are using air,
not 0% O₂ and 0% He (i.e., pure N₂).
So whenever Subsurface deals with Air we do so with a gasmix of all zero.

> Note that this problem is not specific to the Cobalt. I see exactly the same
> with the Mares Darwin Air (and Airlab). That might help for debugging the
> problem, because with the darwin backend you can use the simulator. If you
> disable the 50 ms delay in mares_darwin_device_open, it runs faster :-)

OK, good. That should allow me to test the Subsurface side of things. Not
sure if I'll have the time today because of some family issues, though.

/D