Suunto EON Steel breakages..

[Dirk Hohndel]

Hi Linus,

> On Jun 20, 2016, at 12:03 PM, Linus Torvalds <torvalds at linux-foundation.org> wrote:
> 
> So the fairly recent update of libdivecomputer to merge stuff from
> Jeff is completely broken for the Suunto EON Steel: Jeff added support
> for the "devinfo" event for the EON Steel, but that event is just
> pointless garbage, and actually breaks things by introducing the wrong
> device IDs.
> 
> I don't understand why Jeff refuses to acknowledge that using an
> unsigned int for a serial number is completely broken, and why he
> cannot just take the new string-based interface, but the end result is
> that we now get garbage from libdivecomputer, and sadly then trust the
> garbage we get over the much better model we used to use.

I did a crab job merging things - actually, I should have merged far fewer
commits. This is a side effect of not spending enough time on Subsurface
and trying to quickly bring in the changes for a couple of the newer dive
computers that Jef added in master.

> The end result is that trying to download data from the EON Steel will
> now download old dives as well, since subsurface thinks it's a
> different dive computer.
> 
> I have some preliminary patches to fix it, but they just undo the
> libdivecomputer breakage. I'll try to make soem more invasive changes
> to subsurface to be smarter about device ID's instead, so that we can
> let Jeff continue to play with his useless bogus information, and just
> ignore it when we have better information available.

Thanks, that would be great.

As I mentioned here a few weeks ago, I'd love it if someone else could
take over maintaining the Subsurface-branch of libdivelog. Unfortunately
no one volunteered :-(

> I'll also try to extend the Suunto EON Steel parser to start adding
> the CCR support that I now have a test-case for. We currently don't
> parse setpoints etc correctly, which is not surprising since the
> parser is based on reverse-engineering things based on my dives, that
> didn't have anything like that.

See, there's value in your foolishly risking your life.

> But in the future, I *would* ask that when merging Jeff's
> libdivecomputer branch, not taking the EON Steel changes. All the code
> came from me originally anyway, and so far Jeff's changes have just
> been making things worse.

See above. But yes, I'll be a lot more selective in what I take.
Which is a shame.

/D