Subsurface 4.7 / 5.0

Dirk Hohndel dirk at hohndel.org
Sat Jan 21 06:01:00 PST 2017


On Sat, Jan 21, 2017 at 12:01:40PM +0100, Anton Lundin wrote:
> 
> I'm still tempted to revive the branch where I bult subsurface-core as a
> library with jni wrappers and build a UI in Android native java.
> 
> There are enough competence and tutorials out there so almost anyone can
> hack on a Android java app, and the "core" parts of subsurface, we
> already know.

I am very open to that at this point. The reason to go with QML was the
sales pitch that this would be much easier and there would be much more
code reuse. That is clearly not what happend. instead massive changes were
pushed into Subsurface itself that kept disrupting the development of the
desktop app, in order to make progress for -mobile. And while I have
slowly gotten to theee point where I was able to make sense of QML code
(and I think a major chunk of the mobile app was by now written by me),
not a lot of others have been successful with this.

The downside is that I speak even less Java than I speak QML - so this
approach will put me in a spot I'm rather uncomfortable with in that I
will have to even more rely on others to do the work and fix bugs. But to
be honest, I find it extremely unlikely that this will be worse than what
we have now.

> > Instead, work on fixing the things in Subsurface that are broken,
> > incomplete, or missing. Divesite handling. Statistics. Etc. And add things
> > that matter to people, like support for Bluetooth LE which we will need
> > for a number of new dive computers, starting with the Shearwater Perdix AI
> > that I should receive today and which you can only talk to via BTLE.
> 
> Awesome!

> As far as I've understood, the actual protocol is the same (or at least,
> the Perdix AI supports the Petrel protocol), its just the transport
> thats different.

I have yet to hear back from Shearwater regarding the transport mechanism.
BTLE is not related to BT at all. The shared name is nothing but
marketing. So this will also require quite a bit of work.

> Would be interesting to try to get the AI-data out of it to.

That is of course the goal. But without being able to talk to it, that's
not going to happen.

/D


More information about the subsurface mailing list