Strange duplication of dives

[Jef Driesen]

On 16-12-12 18:40, Dirk Hohndel wrote:
> Jef Driesen <jefdriesen at telenet.be> writes:
>
>>>> @subsurface dev's: Shouldn't the diveid hash that was introduced some time ago
>>>> catch this? The raw timestamp in the dive doesn't change, only the parsed
>>>> date/time value.
>>>
>>> The diveid is simply part of the SHA1 hash of the fingerprint of the
>>> dive. I assume that contains the raw timestamp, correct?
>>
>> For the sensuspro, the libdivecomputer fingerprint is indeed the 4 byte raw
>> timestamp. So the fingerprint (or its hash) is invariant and thus should have
>> triggered the cut off, regardless of whether the final date/time value is
>> different or not.
>
> Hmm, so something is broken there...

I just checked the libdivecomputer side, and seems to be fine, but I could be 
mistaken of course.

>>> We used to cut off downloads by saying "this dive was prior to the last
>>> dive that we already have" - much simpler and certainly prevented what
>>> happened to you...
>>
>> That's not entirely correct. If the duplicate gets a time that is greater than
>> the time of the dive you already have, you would still get a duplicate.
>
> True. But you wouldn't get lots of duplicates, all off by a minute or
> two...

Correct.

Jef