[Talk-us] "DB fixer"?
Richard Fairhurst
richard at systemed.net
Wed May 18 13:31:02 BST 2011
Nathan Edgars II wrote:
> http://www.openstreetmap.org/browse/way/28397519/history
> This way seems to have been reduced to only those nodes at
> intersections, obviously wrong given the curves in the road. Can
> anyone explain what this "DB fixer" is and how much damage
> it's done?
Back in the mists of time, before we had transactions (with API 0.6), it was
possible that you could (IIRC, it's a long time ago) upload a way with P1,
and the server, if under high load, would die halfway through leaving the
data in an inconsistent state. This sometimes meant that ways ended up
containing deleted nodes. IIRC Potlatch 1 and the /way/history call would
show the deleted nodes as part of the way, whereas anything else using the
/map or /way calls (e.g. JOSM) wouldn't.
Stefan de Konink got quite exercised about this and carried out some
automated edits to remove the deleted nodes from the way - rather than
undeleting them, which in my view would have been a better solution.
(Sidenote: Rather depressingly he has since refused ODbL+CT which strikes me
as pretty immoral for automated changes that just delete stuff.)
http://wiki.openstreetmap.org/wiki/OSM_Fixer refers, I think.
The easiest way to restore individual cases is probably to use P1's revert
tool:
- open P1 for the relevant area
- select the way
- press 'H' for history
- go back to a pre-"fixed" version
- click 'Revert'
- with the way selected, press 'J' (join) to squish any dupe nodes
- save
I've just tried this with the case you cited and it seems to work. First
time I've used P1 for ages. ;)
cheers
Richard
--
View this message in context: http://gis.638310.n2.nabble.com/DB-fixer-tp6377376p6377588.html
Sent from the USA mailing list archive at Nabble.com.
More information about the Talk-us
mailing list