[Talk-GB] New wiki page for GB reversion requests

Mark Williams mark.666 at blueyonder.co.uk
Fri Jul 24 00:48:33 BST 2009


Peter Miller wrote:
> On 22 Jul 2009, at 15:18, Andy Allan wrote:
> 
>> On Wed, Jul 22, 2009 at 1:37 PM, Peter Miller<peter.miller at itoworld.com 
>>> wrote:
>>> Without going through every edit in the changeset it will be hard to
>>> determine. If we do have to go through every changeset then we  
>>> might as well
>>> revert them by hand. Possibly we need to leave this until better  
>>> tools are
>>> available or challenge some clever person to write the required  
>>> tool in the
>>> next day or two.
>> OK, I now have a tool that will revert all the components of a
>> changeset that haven't been reverted already, and ignore everything
>> that has been changed since. And now I have a good example of why it's
>> not that straightforward. Take this:
>> http://www.openstreetmap.org/browse/node/242058267/history
>> The guy moved it (v3), and then deleted it (v4). So reverting it would
>> put it back to v2. But if it was deleted out of a way, and that way
>> has been moved since, it wouldn't put it back in the way again since
>> that way wouldn't be reverted. Which makes it a bit pointless. And
>> maybe someone has fixed the way (adding in a new node there, or
>> nearby, or similar) so this node isn't needed. So it's impossible to
>> tell what to do with the node.
>>
>> So after a few hours of investigating this, I'm back to where I
>> started* - reverting changesets is easy so long as nothing has changed
>> since. Anything else needs a graphical editor. Better such tools can
>> be created, and ideas/mockups/code is wanted.
>>
>> So for the future, if there's another changeset that needs sorting
>> out, please consider asking someone to revert it before anyone tries
>> to manually fix it. Manually fixing stuff is of course fine but it's
>> an all-or-nothing approach that can't be "finished off" with a script.
> 
> Ok, so I claimed 6 change-sets with the ones at the top of the list. I  
> checked all the nodes on the first page and then noticed that this was  
> page 1 of 42 of changed nodes - a total of 823 nodes to fix. Now that  
> is 823 nodes (and 39 ways) in one of 35 change-sets. That is  
> potentially a lot of work. Any ideas anyone?
> 

Have played. The changesets are not convenient, as I haven't found a 
good way to load one into JOSM & use it meaningfully.

However I do have a copy of S.E. England before this little pest was 
invented, so I can use that as a background & highlight a given user & 
'blink' layers, which shows up his evilness a treat, including 
deletions. That just leaves tagging changes to find, which again is 
easier with a 'known good' set of data from April. As my part of the 
world was OK back then it's only changed a little in detail, & obviously 
I can see the pests' work distinctly.

At 117Mb it's not quick to load, but very useable.

Mark





More information about the Talk-GB mailing list