[OSM-talk] License plan

[Dave Stubbs]

2009/3/5 SteveC <steve at asklater.com>:
>
> On 27 Feb 2009, at 05:04, Ben Laenen wrote:
>
>> It looks like we finally got some kind of "License plan" for the step
>> towards the new license, so everyone check
>> http://wiki.openstreetmap.org/wiki/Open_Data_License/Implementation_Plan
>>
>> Let me start with the obvious questions first:
>>
>>
>>
>> * why don't you split between the votes whether you like license X and
>> the question whether you're allowing the change of license on your
>> data?
>>
>> After all, I want to have an idea *how much* of the data will still be
>> there after the second vote. If it turns out that any data from
>> someone
>> who gave his approval would be deleted, then count me as no vote.
>
> so vote method is an interesting constraint... but I think we're being
> really hardcore in making sure that everyone who added data has to
> agree or we reset the process back to zero.
>
>> * I still have no response to the question what would happen with my
>> data if it's derived from someone who doesn't give it's approval for a
>> license change.
>
> My view, personally, is that it should be dropped. But y'know I just
> don't think it will happen like that. If we build a positive process
> and bring people with us then we'll get the majority of the people
> along. We will lose small bits of data but thats ok, we have fantastic
> volunteer community to fix those edges, we'll be in shape in no time.
>
>> And how are you going to check that anyway? You can do lots of things
>> with CC-BY-SA data (copying, splitting, merging) where it's impossible
>> to
>
> Well it's all in the database... every single edit (oh and the dump of
> the segments stuff). Because one day, about 4.5 years ago I knew it
> would be needed and designed it in. No need to thank me. No. Really.
>

I think he's referring to the more complex derivatives (such as
splitting a way) which are clearly logically derived but have no
obvious connection in our data model (splitting creates a new way with
no history, as well as leaving the original way in place but
truncated). But this is only an issue if you're taking the literal
unwind objects only approach and still regard such artefacts as in any
way significant. Mostly nodes will get in the way anyway and force you
to do some kind of cascade delete, but there are still more than a few
ways these bits and bobs will leak through.


>> My understanding is for example that if you split a way, there's not a
>> single connection between the two parts of the way telling that one
>> derived from the other.
>
> I can't comment on what potlatch does to my beautiful database.
>

You can however comment on your crappy data model that doesn't allow
forking an object to share a common history :-)

Dave