Hi,
the German forum came up with a nice example of what can happen if
import scripts don't work as expected:
http://www.openstreetmap.org/browse/node/1225280729
- a node that is part of over 200 ways ;)
Trying to show these ways on the API "browse" page results in an error
500, probably because the API tries to tell us for each of the 100 nodes
which 200 other ways they are also a member of...
The interesting lesson to learn here is:
* This import *was* actually properly documented
(http://wiki.openstreetmap.org/w/index.php?title=Australia/Queensland/The_Department_of_Environment_and_Resource_Management/Protected_Areas_Import)
* The file that was proposed for uploading *was* available for review
(http://dl.dropbox.com/u/446994/simplifiedQldParks.osm.zip).
* The 52 MB file *did* already contain the errors - one node in the file
was even referenced by 572 ways, a total of 578 nodes referenced by 10
or more ways -, but the problem was not spotted.
It seems that JOSM was used for uploading this, and today's JOSM
validator would have spotted the problem (it reports 9 duplicated ways,
122 ways with same position, 287 self-intersecting ways and many more),
but it is possible that back in 2011 when this was made, the validator
was less advanced or more easily overlooked.
When thinking about improved import guidelines, maybe we should not only
request that people publish what they want to import, we should also ask
for one or two mappers with a certain track record to "second" the
import, something like "I have examined the proposed data and I think it
is fit for uploading, signed, John Doe". If there is an imports group
like the one that Serge is setting up in the US, that group would
routinely "sign off" an import like that, but anyone else who knows what
they're soing could do so too.
Bye
Frederik
--
Frederik Ramm ## eMail [email protected] ## N49°00'09" E008°23'33"
_______________________________________________
Imports mailing list
[email protected]
http://lists.openstreetmap.org/listinfo/imports
