[OSM-talk] API anomalies (was: Potlatch and merging ways)

Matthias Julius lists at julius-net.net
Tue Jul 3 21:33:22 BST 2007

Matthias Julius <lists at julius-net.net> writes:

> I think a pretty nasty bug has appeared in Potlatch a couple of days
> ago.  I had it happen to me a couple of times now that when merging
> two ways (by shift-click on an endpoint while drawing) the second way
> gets deleted as it should, but the first one does not get extended.
> When downloading the data into JOSM the second half is just not
> there while it is still shown in Potlatch.  I then tried to insert
> another node into the missing part of the way.  This node gets
> uploaded and does show up in JOSM, but the way is still incomplete.
> When I try to delete that node in JOSM I get error 412 "Precondition
> failed".
> That's how far I got with my investigation so far.

I created a test way (4850696).  Then I checked its history.

This is the last way element of 

<way id="4850696" visible="true" timestamp="2007-07-03T21:06:09+01:00">
  <seg id="27418663"/>
  <seg id="27418720"/>
  <seg id="27418721"/>
  <seg id="27418722"/>
  <tag k="created_by" v="Potlatch alpha"/>

And this is from the output of 

<way id="4850696" visible="true" timestamp="2007-07-03T21:06:09+01:00">
  <seg id="27418663"/>
  <tag k="created_by" v="Potlatch alpha"/>

Shouldn't that be the same?  How is that possible?

The segments except the first one look all like this

<segment id="27418722" from="31267124" to="31267001" visible="true" timestamp="2007-07-03T21:06:09+01:00"/>

They are all visible="true" but have no nodes.

There is something fishy.


More information about the talk mailing list