I posted the original question/comment and have been testing a switch over to
providing source information _only_ on changesets. I posted a followup message
which related the source-on-changeset matter to provenance in RDF data .. I
think that the source-on-changeset is as valid as and more conservative with
respect to data duplication than the source-on-object solution. I won't go
into detail about RDF quads and quints versus linked data-based provenance
because it makes my head hurt to just think about it (I'm early in the learning
curve), but the source-on-changeset is a more elegant solution than the
source-on-object and what we need to do is to revise the Potlatch and JOSM
interfaces in a way to take advantage of this elegance and expose the source
data to editors in a similar manner regardless of whether it is on the object
or the changeset.
>From an editing point of view, leaving source to be reported on changeset
>works to significantly reduce the amount of tag-value content you have to deal
>with while adding content. That is a very good thing.
In closing -- I'm a convert to source, source_ref, source:date and other
variations being added to the changeset ... at least given my current editing
behavior.
--ceyockey
-----Original Message-----
>From: Ilari Kajaste <ilari.kaja...@iki.fi>
>Sent: Jan 6, 2013 6:06 PM
>To: "Tag discussion, strategy and related tools" <tagging@openstreetmap.org>
>Subject: Re: [Tagging] Source tag - deprecated for use on objects?
>
>(Replying to the entire conversation thread)
>
>The way I see it: No data about the objects themselves should be in
>changeset description - changeset should only describe the *change*
>itself. Because - at least to me, but I do expect to many others as
>well - it makes the database structure more clear and intuitive.
>
>Even though source tag is metadata, it's still data about the objects.
>
>I sort of agree that metadata doesn't belong on objects. Or maybe it
>should at least have some header differentiating it from actual data
>("meta:source"). But having data about objects stored on changesets is
>still worse.
>
>Source tag does indeed have described problems of specificity
>("bing;survey;knowledge" - which is for which tag) - but these are
>mostly edge cases, and a lot of other things in OSM have very similar
>problems, and still manage to be very useful in most cases. What needs
>to be kept in mind is the *usual* cases where the information is
>useful: when I see a feature on OSM that conflicts with what I would
>edit there, source tag gives me at least some understanding on which
>information is more correct. E.g. when correcting a way to match Bing
>imagery or GPS trail, "source=landsat" is a very different case to
>"source=survey". Or when changing a name, "source=knowledge" is very
>different from no source tag at all.
>
>Disclaimer: Not attempting to force my way of thinking on anyone, just
>providing my viewpoint on the subject for the discussion.
>
>On 2 January 2013 14:50, <dies38...@mypacks.net> wrote:
>> I have been told ( on the talk-US email list ) that use of {{key|source}} on
>> objects has been deprecated for years and that such information is only of
>> historical interest and its use should be restricted to changesets.
>
>--
> - Ilari Kajaste -
>
>E-Mail: ilari.kaja...@iki.fi
>WWW: http://iki.fi/ilari.kajaste
>
>_______________________________________________
>Tagging mailing list
>Tagging@openstreetmap.org
>http://lists.openstreetmap.org/listinfo/tagging
_______________________________________________
Tagging mailing list
Tagging@openstreetmap.org
http://lists.openstreetmap.org/listinfo/tagging
