Strange duplication of dives

Jef Driesen jefdriesen at telenet.be
Sun Dec 16 12:26:42 PST 2012


On 16-12-12 19:11, Linus Torvalds wrote:
> In reality, that's generally what you want: when you download new
> dives, they'll obviously be newer than the old ones, and the trip
> merging will work fine.
>
> The problem here is that the dive matching didn't work due to the
> insane sensus date problem (and it has *never* worked - Jef has talked
> about the issue, but the 1-minute fuzz has apparently caught it for
> you before), and then from that, the end result is this kind of "once
> insane dives happen, we create insane automatic trips for them".

I'm not sure you aware of the fact that the Reefnet devices (all three models) 
are headless data recorders. Unlike real dive computer, they have no user 
interface at all, and thus their time keeping method does make sense.

Anyway, I think the right model is to not download new dives at all. Then no 
merging is necessary. The best method (if you ask me :-)) is to rely on the 
libdivecomputer fingerprint feature, or implement it in the application. But in 
the latter case you should make sure to rely on some fingerprint like info (e.g 
the subsurface diveid). Certainly not the *parsed* dive timestamp!

But as you mentioned already in one of the other replies, in this case the root 
problem is likely that the old dive didn't have a diveid yet.

Jef


More information about the subsurface mailing list