On 9 Nov 2010, at 02:47, Serge Wroclawski wrote: > On Mon, Nov 8, 2010 at 7:23 PM, Richard Palmer >> I'm interested in adding historical information to the OSM database ...It's >> been pointed out to me that there was a similiar proposal put by Frankie >> Roberto some time ago; I wondered if any conclusion was come to about adding >> this data to the main database? > > I don't understand the question. Why can't you use the minutely > planet.osm updates to construct this database?
Serge, you misunderstand the question. An example of what Richard suggests would be the mapping of the Crystal Palace in London. This was a very large iron and glass structure that opened in Hyde Park in 1851, was relocated to Sydenham (7.5 miles away) in 1854, then burned down in 1936. http://en.wikipedia.org/wiki/The_Crystal_Palace Using start_date and end_date tags on two way IDs (one for the Palace in its first position, another for its second) might be a reasonable way to map the Crystal Palace. The question is: is the main OSM database the best place for this? Some specific issues (off the top of my head): * Suspect data (inaccurate or copyright-tainted) cannot be resurveyed. * Every normal renderer (i.e. one that's aiming for the most up-to-date rendering) has to check for an end_date tag on every node and way. * Every tool writer has to implement a time filter and face demands for a time UI. * Does OSM's existing concept of "history" sit well with additions of historical data? * No obvious way to handle extensions to and partial demolitions of buildings (these are frequent in many important buildings), nor to represent continuity in buildings that are moved (not just Crystal Palace... [1]). And some broader ones: * Would OSM's database size increase be worth the payoff? * Would OSM's database complexity increase be worth the payoff? * What is that size increase? * What is that complexity increase? * Would the expansion in OSM's apparent mission ("all of history") be beneficial to the project? - L [1] http://www.youtube.com/watch?v=Zz4ehBCF8a4 _______________________________________________ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk