[OSM-talk] Data oddness

Richard Fairhurst richard at systemed.net
Mon Dec 8 17:19:16 GMT 2008

D Tucny wrote:
> The lack of history of a previous version would suggest that either 
> there was no previous version and the way was created with existing 
> nodes, or that potlatch removed the previous way and replaced 
> it with a new way which due to some communication issue then 
> potentially showed as if the nodes were POIs which the user 
> then deleted after the fact and for whatever reason allowed the 
> way to be corrupted...

Shouldn't be the latter: amf_controller (the bit of the API that Potlatch
talks to) won't let you delete a POI that's become part of a way. But as
frequently posted, in a transaction-less database this kind of inconsistency
can and will happen. In my experience it's usually splitting or merging ways
at times of high server load.

> The fix would probably be to undelete the individual nodes that 
> make up the way...

...and you can do that just by getting Potlatch to write the whole way
again. Select the way, do something to it (anything you like, just change a
tag or move a node a short way or something), and deselect. When it's
rewritten all the nodes will be made visible again.

View this message in context: http://www.nabble.com/Data-oddness-tp20893679p20899554.html
Sent from the OpenStreetMap - General mailing list archive at Nabble.com.

More information about the talk mailing list