[Rebuild] Example of a questionable changeset - will it be automatically reverted?

[Kevin Halton]

Hi,
Sorry I'm not very familiar with the rebuild strategy, but I know there are several algorithms being used to determine what must be deleted, what reverted and what can stay.
I have seen several changesets recently with descriptions similar to "relicensing" or "licence change", where it appears that the nodes are just being copied and pasted into new ways.  I would like to know if there is an established approach to such changesets, and whether they will be automatically identified and reverted?
 
Here's one example, from the Czech Republic:
http://www.openstreetmap.org/browse/changeset/11286636
 
It was made along with dozens of other very similar changesets, by the same person, and following a list of way ids in need of "replacement" which was recently published.  Some of these changesets follow each other in 1- or 2-minute intervals, indicating that whatever method was used, it was quick.
 
In this changeset, one way was deleted, and most of its nodes also, and a new way was created, with new nodes.  Here are two of the nodes, before and after:
Before:
node id="75059253" lat="49.401853" lon="13.341798" user="pavel"
After:
node id="1714266285" lat="49.401853" lon="13.341798" user="WernerP"

Before:
node id="75067886" lat="49.402566" lon="13.343281" user="pavel"
After:
node id="1714266286" lat="49.402566" lon="13.343281" user="WernerP"

Before:
node id="75045319" lat="49.403673" lon="13.344304" user="pavel"
After:
node id="1714266292" lat="49.403673" lon="13.344304" user="WernerP"
 
This means to me, that the "clean" status has been edited, but the coordinates are identical.  In my view the data is still not clean and the whole changeset should be reverted.  I presume that the other changesets in the series have similar qualities (although I have only looked at two of them so far).  My first question to this list is then, do you agree that this changeset is questionable or does it comply with the cleanup guidelines?
 
If it should be reverted, then my next question would be, can the redaction bot spot this kind of problem automatically or not?  In this case the delete and paste are in the same changeset so I assume with some work it might be possible.  If they're in different changesets then I guess not.  Will this changeset be handled by the redaction bot as it is, or would that require additional work?
 
Thanks,
Kevin
  		 	   		  
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/rebuild/attachments/20120416/d586437c/attachment.html>