[OSM-talk] Wanted feature for API 0.7 ??

[Matt Amos]

2009/6/8 Peter Dörrie <peter.doerrie at googlemail.com>:
> 2009/6/8 Frederik Ramm <frederik at remote.org>
>> There are many unsolved questions here. For example: What happens if parts
>> of the "ancient" world transcend your "fourth dimension", e.g. a
>> contemporary secondary road uses a few bits of an ancient Roman road. They
>> would surely share the same nodes, wouldn't they? But if someone then
>> deletes the secondary road (which he downloaded without ever knowing that
>> the Roman road also exists because that was shielded from him), he must not
>> delete the nodes because they are still used by other objects...
>
> My angle on this is primarily the historical-genetic one. Taking your
> example:

i'm pretty sure this isn't the example that frederik was talking about.

> 1. Brutus Mappus maps this roman long distance road in 100 B.C he tags it
> correctly with highway=roman and surface=cobblestone. The road is used in
> that form for the better part of the next two millennia.
>
> 2. In the 19th century it gets some heavy usage and deteriorates. The local
> government decides to build a new road, which uses some of the same vectors
> the roman road used so far. The grat-great-great (etc) son of Mappus (John
> Maps) splits the road, tagging part of it as highway=disused and others as
> highway=construction.

and if he just deletes the road?

> 3. The new highway is ready and John tags it as highway=primary
>
> 4. The same thing happens several times over until fake Steve finds a
> motorway, using some parts of the primary road, using some parts of the
> roman road (which by now is not longer visible in the landscape and has been
> tagged as historic).

until Victor Vandal comes along and moves some of those nodes halfway
across the world, or uses them to spell a rude word.

> Okay, what does this mean for whom?
>
> Users: The normal user will see a rendering which shows "what is on the
> ground". -> Motorway and those parts of the primary road that still exist
>
> Mappers: The normal mapper will see, what is relevant to him. -> Same as
> user and some additional tags (oneway, surface, maxspeed, etc.) plus perhaps
> those "disused" roads, as they may still be relevant to mapping.
>
> Special interest person (scientist, etc): He gets the possibility of seeing
> exactly what he wants to see. The situation in 100 B.C? no problem. Ask the
> database about all disused / historic / etc objects? also no problem.
>
> Different object sharing the same nodes over time (and changing them) is not
> a problem.

this "4-D" usage of OSM has some problems:

1) given enough eyeballs, all bugs are shallow. in other words, we
shouldn't be hiding data; the more users who are looking at the data,
the more users who are correcting and extending it. any "special
interest" layer would, by definition, be default hidden.

2) sharing information which shouldn't be shared. there will be two
ways from different "special interest" layers which share a node when
they shouldn't, leading to all sorts of fun when that someone tries to
delete that node from one layer and can't because it's being used by a
way from an invisible layer. this was frederik's point, i think.

3) not sharing information which should be shared. if i went to a
roman road and surveyed it extremely accurately and, because its a
currently-in-use road, lined up the current highway with the
measurements that would update the "special interest" roman road,
right? well, it depends on how that works. if the roman road and
current road share all their nodes, but i add a node to the current
road, does that make the client add a node to the roman road too?
again, the client can't know this without downloading all the data.

my solution would be to set up my own OSM server for this kind of
purely historical data, where it won't conflict with existing OSM
data.

cheers,

matt