cloud connection timeout

[Dirk Hohndel]

> On Mar 22, 2016, at 5:33 PM, Miika Turkia <miika.turkia at gmail.com> wrote:
> 
>> On 22 Mar 2016, at 22:16, Dirk Hohndel <dirk at hohndel.org> wrote:
>> 
>> 
>>> On Mar 21, 2016, at 5:17 PM, Miika Turkia <miika.turkia at gmail.com> wrote:
>>> 
>>> I am currently suffering from poor network connectivity and cannot sync the desktop app to the cloud. Looks like it times out before ssl hand shake is finished. I tested the connectivity with openssl and it took about 60 seconds for the connection to initialize properly to cloud port 443. (TCP connects within a second or two.) Can the timeout be easily extended somehow? As we all know many divers do frequent this kind of poorly networked locations.
>>> 
>>> I can grab verbose logs or network capture at some point if my initial assumption of the root cause proofs to be wrong.
>> 
>> If openssl takes 60 seconds, then Subsurface won't be any faster - and we time out way before that.
>> 
>> I have thought about this problem (and I'm pretty sure we discussed it here) before. The problem is that once we successfully connect with the server we reset the dive list - which means that changes that the user has made since we initiated the cloud connection might get lost. If we allow massively delayed sync (a minute or more), then this is likely to cause users some frustration.
>> 
>> We could freeze the UI or make things read only while the sync is running, but I'd perceive that as a really poor user experience, especially if I'm at a place where the sync won't complete, ever.
>> 
>> I'm open to ideas - but simply making the time out longer would seem to be asking for trouble.
> 
> A dialog asking whether to attempt slow sync mode or work off-line comes to mind.
> 
> I just managed to run the sync and even now the ui was totally frozen for several minutes during the sync. Of course a different case as git was working...

So in all honesty - I'm not sure what the right answer is here. I don't think that a "several minutes" sync time is useful at all. That's why we have the local cache.
The problem is... how do we get people to do the first sync where they have decent bandwidth and not on a trip...

/D