UEMIS Patch fixing alternating dive details

[Linus Torvalds]

On Sat, Sep 19, 2015 at 5:15 PM, Linus Torvalds
<torvalds at linux-foundation.org> wrote:
>
> This is *NOT*TESTED*. It's still downloading without this patch (and
> with the manual clearing), and I wanted to let that go on.

Ok, finally tested. The downloading took forever, but part of it was
that since it needs manual intervention for the unplug/replug, and I
was doing other things too, it took literally hours.

Anyway, the patch seems to work for me. The "force download all dives"
worked, and it also did the right thing when I deleted a few recent
dives and tried to just do the normal incremental thing.

That said, I had some really funky behavior when I compared the
results of the downloads. The first time looked *slightly* different,
but the differences seem to be in how it downloaded the same first few
dives twice (due to the original "force" bug), and then odd things
happened when merging what should be identical dives. Like the
beginning time being off by a second. How the heck does that happen?

I also had a strange duplicate dive fro my insane 12-hour dive -
that's when my uemis decided to enter dive-mode when descending in a
plane, and was convinced I was diving when at ground level. It
actually did that for much longer than 12 hours, but the Uemis seems
unable to record that. But again, I don't think my patch had anything
to do with the difference, it's more likely either due to the original
bug, or a sign of some other unrelated oddity in the downloader.

BUT.

I think my patch is fine. The problems I saw were either due to the
crazy duplication that you get without my patch, _or_ due to just
generic uemis downloader problems.

It should still get some more testing and checking by others.

                            Linus