[OSM-talk] Can I say "yes" to the ODbL if I can't account for 100% of my data?

Toby Murray toby.murray at gmail.com
Thu Jun 16 19:03:04 BST 2011


On Thu, Jun 16, 2011 at 12:50 PM, SomeoneElse
<lists at mail.atownsend.org.uk> wrote:
> On 16/06/2011 18:00, M∡rtin Koppenhoefer wrote:
>>
>> You can also put this information in the change-set-comment. IMHO this
>> is where this belongs to. AFAIK the source-tag is disputed and it is
>> recommended to use the changeset comments.
>
> The problem with the changeset "source" tag is that there's no granularity -
> one tag applies to the whole edit. Presumably the only time that this would
> be valid would be an entirely armchair tracing session with no local
> knowledge and no other on-the-ground evidence (surely not recommended) or an
> import (which should surely afterwards be tidied up with local knowledge
> anyway).
>
> Using changeset comments is even worse; it's just a bunch of text associated
> with a particular edit. If I wanted to know the source for updates for the
> Pennine Way (http://www.openstreetmap.org/browse/relation/63872) I'd need to
> fetch the relation history (not feasible via the API I suspect), fetch 209
> changeset details, and manually parse a lump of English text in each one for
> something that might be a "source".

I also frequently use multiple sources in a single changeset. I often
flip between faster bing imagery and slower to load but better aligned
and sometimes more distinct NAIP imagery. Which one I actually trace a
feature from depends on multiple factors. Hence I often put source
tags on individual objects. And with 84 million of them in the
database, I would hardly call this tag "disputed"

Toby



More information about the talk mailing list