regressions because of the new tank handling

[Linus Torvalds]

On Thu, Sep 14, 2017 at 11:05 AM, Linus Torvalds
<torvalds at linux-foundation.org> wrote:
>
> Ok, I found a real bug, it wasn't me fat-fingering the XML.
>
> That bug has nothing to do with the problem Stefan sees, but the
> multi-sensor pressure saving is definitely buggy for XML. I'll send a
> patch.

Fix sent.

Interestingly, there's another bug there in how we interpolate the
pressures that also only seems to show with that test-case.

That one is less obvious, and I think only affects the display of the
pressure graph (because it's about our interpolation between pressure
samples)

I haven't figured out the cause yet, but the reason I hadn't noticed
it before is the same "this is a planned dive with samples far away
from each other". That makes the interpolation bug show up in ways it
didn't for any of the dives that have dense pressure samples.

                 Linus