#772

[Dirk Hohndel]

On Mon, Nov 24, 2014 at 06:06:36PM +0200, Willem Ferguson wrote:
> >>to my (non-existent) knowledge if one adds cylinders to a dive, those
> >>will remain unused and cannot be marked as used and participate in the
> >>calculations. is that correct?
> >I just commented on the bug. I have added support for this scenario a
> >while ago. Once you manually add tank switches between the different tanks
> >with identical gas, things should work correctly.
> 
> Please give a little bit more information, both for myself and for the user
> manual. Attached is the profile of a recreational sidemount dive. Unless I
> let the gas mixture differ by 2%, the gas switches are not recorded. So I
> get round this by using gas mixtures 1% above and 1% below the true mixture.
> Since both cylinders almost never have exactly the same O2, the mixtures on
> the graph are frequently very close to the measurements at the start of the
> dive. But it is still a workaround. If there is a way to switch cylinders
> with exactly the same mixture, it would be very useful to know how to
> achieve this and I would like to document that in the user manual.

So since commit 669da22d8cc8 we have the ability to explicitly switch to a
specific tank, even if multiple tanks have the same gasmix. Linus actually
implemented the necessary infrastructure a couple months earlier, but
above commit adds the first UI to do this as a user (and I improved on
this UI in the past week).

The idea is this: as a side mount dive (or cave diver, or whatever) with
multiple tanks with the same gas, you can add the gas switches via the
context menu on the profile. If you also have starting and end pressure of
those tanks, it even does the SAC calculations correctly.

There are a couple of pitfalls: many dive computers simply don't allow you
to have multiple tanks with the same gas (e.g. the Petrel). And as of
today we only get switches to a GAS, not to a TANK from libdivecomputer.

The only dive computer where we have full support for this is the Uemis.
I'll need sample dives from other dive computers to figure out how to
return the necessary information from libdivecomputer. Jef is adding the
APIs to do this in 0.5, so I think it's now possible.

Let me know if this sufficiently explains the situation

/D