[OSM-talk] The Fourth Dimension (temporary mapping)

Frederik Ramm frederik at remote.org
Tue Jul 24 12:54:38 BST 2007


    reading about the floods in England prompted me to think about  
whether we should make efforts to somehow include "temporary mapping"  
for OSM.

There are many aspects of maps that change "temporarily". Some roads  
are closed in winter, but they are not closed at a certain date -  
they are closed "in wintry conditions". Some roads are closed long- 
term (construction, bridge or tunnel refurbishing), some short-term  
(cleanup after accident). Lakes dry up, and flooding creates "lakes"  
where there should be villages.

Some of these changes are fixed in start and end date; some just  
happen, and are gone after a certain time.

Now we can of course make "mashups" of any kind - "pins on maps"  
being the most primitive, or the better way that OJW recently  
presented where you can draw on top of maps, or other things maybe.

However, all these are based on having a fixed, raster data base map  
and somehow marking stuff on top of that. Which is way inferior to  
drawing proper maps from vector data.

If we are to enter time-limited data int our database, this will have  
a lot of consequences. For example, we will need at least two types  
of rendered maps - one "classic" map showing the "normative" state of  
things, how things "usually are", maps that you would use as base  
maps for drawing on. And another one showing the "current" state of  
things, with lakes removed when dry, villages submerged, roads closed  
and so on.

We also need to find proper tags or change the API to model these  
things. We could just set "valid-from" and "valid-to" tags and have  
renderers etc. deal with it, but (thinking that we might some day  
also have ancient roman roads and stuff) it may be an option have a  
query whereby you can say "give me this bounding box but only stuff  
that's valid for August 2007" or so.

I don't think that our existing history mechanism is sufficient - I  
can access the status quo for a certain date in the past, but there's  
no way to enter information like "this road will be closed on 29th  
July and subsequently demolished" (I'd either have to set up a cron  
job or put a "note" in hoping that someone who sees the road after  
29th July will then delete it.)

All this is not urgent stuff we need to solve but interesting to  
think about. - Maybe historic and future data as well as short-term  
changes are not even "our business" and we'd rather count on someone  
else to use our data and modify it according to current developments.


Frederik Ramm  ##  eMail frederik at remote.org  ##  N49°00.09' E008°23.33'

More information about the talk mailing list