On Thu, Apr 11, 2019 at 11:13:35AM +0200, Robert Helling wrote:
> 
> this is really a tough one. Let’s keep in mind in which direction this (if it 
> takes off at all) might go to: Rather than sharing your own dive sites 
> between a few of your own log files (or maybe with your wife or your close 
> friends), this whole dive site management becomes really powerful as soon as 
> some big dive site collections are created (like all dive sites in a country, 
> a continent or even globally) with the collections shared essentially between 
> all Subsurface users. This would allow you pretty much automatically fill in 
> dive site data based only on GPS since pretty sure, somebody will have dived 
> the same dive site at some point in the past. These collections will contain 
> hundreds or thousands of dive sites so manual import selection is not viable. 
> These collection could also be used to figure out where it is actually worth 
> going to. Another aspect is of course language: We might want to remember the 
> langue of a dive site description so that I can decide to only import those 
> that are in a language that I can understand.

I'll admit that I hadn't intended to create a new social network. Nor had
I contemplated building the back-end service that supports all this, but
if this is where people want to take it, I'm certainly willing to
entertain the notion.

From a design perspective that means that any text we store needs to
tagged with the corresponding language string - otherwise how would you be
able to maintain translations into different languages, right? Because for
my own dive log I pick the language I care about, but if I want to be able
to share dive sites outside of my immediate circle of friends and family,
that consistency in language becomes less obvious.

This impacts a lot of things from keeping that information at runtime to
data format at rest. One obvious first simplification to make is to design
in a backwards compatibility to our likely initial state: if no language
information is stored, it's assumed to be in the language that Subsurface
is running in. That (I think) would allow us to release a 4.9 version that
doesn't track language, yet.

Somebody please check my logic here :-)

> I am not saying that we need to implement all this right now, just keep in 
> mind that this is a direction into which things might develop.
> 
> Regarding identifying dive sites that are in some sense close enough: This is 
> not just a matter of comparing two sites: Most of my local dives are in lake 
> Starnberg: https://goo.gl/maps/msX7Mv29mz72 along a roughly 1 km stretch on 
> the east coast. In principle you can enter the water at almost any place 
> along that stretch but there are definitely distinct dive sites, I can think 
> of at least five names people use for different entry points with dives 
> hardly ever overlapping. Would you want to group all into one single site as 
> any two GPS fixes of entry points are connected by other entry points which 
> are pairwise close by? Or do you want to treat those as all different?
> 
> In the current system, I have five dive sites that I refer to by name.  But 
> as soon as I start importing other people’s dive site collections they will 
> use different names. So what shall I do?

I think the logic proposed by Doug and others makes sense. If you have
your own named dive sites, then those should be preferred. If you are
searching for a dive site near where your dive was, then we should show a
list of those within N meters of the location (that either comes from your
GPS info or from something that you picked on a map) should be shown on
the map and you can decide which one to pick. Would that solve your
problem?

/D

_______________________________________________
subsurface mailing list
[email protected]
http://lists.subsurface-divelog.org/cgi-bin/mailman/listinfo/subsurface

Reply via email to