Few bugs and a feature req from testing Subsurface

[Jef Driesen]

On 2014-03-14 00:26, Dirk Hohndel wrote:
> On Thu, 2014-03-13 at 23:06 +0200, Miika Turkia wrote:
> 
>> So here is log of the two dives. Seems that Vyper, that was downloaded
>> first, does not have any gas change events on it. On the other hand
>> Stinger has one such event...
>> 
>> Shouldn't a gaschange event be generated on download even if the
>> computer does not have such? This dive is the only one with 1
>> gaschange event among the 35 downloaded from my two computers, all the
>> other dives are without one. Do we only generate gaschange events when
>> diving air?
> 
> As always, it's complicated. For Suunto divecomputers we should have
> exactly the gas change events that the computer sends to us. On multi
> gas capable Suuntos you ALWAYS start with the first gas, so there's no
> change necessary.
> 
> On some other dive computers you can start with a different gas. And
> depending on how the DC works, some times we end up encoding that
> information with an artificial gas change event at the very start of 
> the
> dive.
> 
> I literally would have to read through the code why this event suddenly
> shows up. Is it possible that you had the Vyper Air set for EAN32 and
> the Stinger for Air? That might confuse things...

Unfortunately, it's even more complex. It turns out that at least for 
some Suunto's the initial gas mix (e.g. the one you start your dive 
with) is not necessary the first one. I recently got a bug report [1] 
for the Helo2, and this model seems store the index of the initial gas 
in the header. Other models likely do the same, but so far I haven't 
been able to figure this out for other models.

Related to the same bugreport, I'm trying to fix libdivecomputer to 
always generate a gaschange event on the first sample, to indicate the 
initial gas mix explicitly. Some backend already do this, but not all.

[1] http://trac.libdivecomputer.org/ticket/2

Jef