deviceid

[Anton Lundin]

On 20 November, 2014 - Dirk Hohndel wrote:

> So Anton brought this up on trac. And I have been bitten by this myself...
> 
> Linus, this code has been around forever, I think this was a very early
> design decision that you made:
> 
> devdata->deviceid = calculate_sha1(devinfo->model, devinfo->firmware, serial);
> 
> Can you remind us, why the firmware version was included in this?
> 
> The problem is, of course, that with every firmware update (and those are
> becoming much more frequent, as you are seeing yourself), you get a new
> deviceid and we redownload all dives :-(
> 
> Now of course we can't simply change this, because then EVERYONE gets to
> download everything again... So I wonder if there's some scheme that would
> allow us not to fail for the common case, but also to get things right in
> the future.
> 
> The only idea I came up with so far is to add a new field
> 
> deviceid2 = calculate_sha(devinfo->model, serial);
> 
> If an existing dive doesn't have a deviceid2 stored, we compare with the
> old deviceid (so if the firmware didn't change, all is well). And for all
> newly downloaded dives, we store both numbers - and if deviceid2 is
> available we compare that.
> 
> What am I missing with this idea?
> 

I wrote it down in trac so i wouldn't forget it and just get annoyed the
next time i run into it...

One quick hack could be to calculate a "new" version of the deviceid,
and if no such exists, calculate a "old" style deviceid and see if that
gets you a divecomputer. If so, change that deviceid to the new-style
and you're migrated.

It will mess up switching back to older versions of Subsurface and
download dives from your divecomputer, but hey, who actually downgrades?
...


//Anton


-- 
Anton Lundin	+46702-161604