[OSM-dev] Possible Potlatch Bugs (was: Daily Diff Merging)

[Brett Henderson]

There are 26675 referential integrity issues in my snapshot-20071019.osm 
file but I haven't found any caused by bugs in osmosis. They are 
available in the following report:
http://www.bretth.com/osmosis/integrity-report-071019.csv.gz

17312 of them are ways referring to nodes with id 0. The most recent of 
these (way 8171053) appears to have been created by potlatch on 
2007-10-06T19:45:21+01:00. I don't know if this bug has been fixed but 
that way is two weeks ago which is a good sign.

The latest error in the report is way 9832391 referring to non-existent 
node 78597899 and checked their history.
http://www.openstreetmap.com/api/0.5/way/9832391/history
http://www.openstreetmap.com/api/0.5/node/78597899/history

The node history is:
<node id="78597899" lat="31.2340263" lon="121.484451" user="Piet 
Stevens" visible="true" timestamp="2007-10-18T22:40:00+01:00"/>
<node id="78597899" lat="31.2340263" lon="121.484451" user="Piet 
Stevens" visible="false" timestamp="2007-10-18T22:41:04+01:00"/>
<node id="78597899" lat="31.2340263" lon="121.484451" user="Piet 
Stevens" visible="false" timestamp="2007-10-18T22:41:10+01:00"/>

The way history is:
<way id="9832391" visible="true" timestamp="2007-10-18T22:41:06+01:00" 
user="Piet Stevens">
<nd id="78597891"/>
<nd id="78597892"/>
<nd id="78597893"/>
<nd id="78597894"/>
<nd id="78597896"/>
<nd id="78597898"/>
<nd id="78597899"/>
<tag k="created_by" v="Potlatch alpha"/>
</way>
<way id="9832391" visible="true" timestamp="2007-10-18T22:41:10+01:00" 
user="Piet Stevens">
<nd id="78597891"/>
<nd id="78597892"/>
<nd id="78597893"/>
<nd id="78597894"/>
<nd id="78597896"/>
<nd id="78597898"/>
<nd id="78597899"/>
<tag k="created_by" v="Potlatch alpha"/>
<tag k="highway" v="primary"/>
</way>

The history is a bit odd. The sequence of events appears to be:
* Node is created.
* Node is deleted immediately followed by way creation referring to it.
* Way and node are both updated but node is still deleted (ie. not visible).

Can anybody shed some light on why this is occurring?

Brett Henderson wrote:
>
> In theory, there shouldn't be any referential integrity issues due to 
> the planet creation process, I'll run a report to test it out.
>