[Rebuild] Redaction bot plugin for JOSM (was: Do we need a singe run bot?)

[Hendrik Oesterlin]

"Simon Legner" simon.legner at gmail.com wrote on 05/06/2012 at 04:39:56 +1100
subject "[Rebuild] Do we need a singe run bot?" :

> On 04/06/12 01:16, Hendrik Oesterlin wrote:

>> If it would be possible to have a plugin in JOSM calling the redaction
>> bot algorithms and doing the required redaction, the human contributor
>> could immediately see if damage has occurred and correct it strait
>> away.

> In order to determine the ODbL status of an object, its entire history
> is needed. The API does not provide a reasonably fast way to access the
> history of several objects.

> For this reason, the licensechange plugin relies on the Quick History 
> Service
>    http://wtfe.gryph.de/
> which provides a quick access to already analysed data.

> The suggested redaction plugin would need access to a similar service to
> achieve a reasonable performance. However, I'm not sure whether someone
> volunteers for the development of such a service+plugin (when the 
> license change should already be completed according to the preliminary
> schedule).

True, the licence change shedule suggested that the work should
already be done since April, 1

But it is obvious that things are not as easy.

Is is also obvious that hug damage will occur and applications
requiring good data consistency and quality as routing in GPS devices
will be broken for quite a long time. There are lots of regions only
mapped by tourists never coming back to do maintenance. It will
certainly be wise for such applications to stay with pre-bot data for a
couple of month or even longer.

The actually available tools to perform remapping and licence cleanup
are not very performant. There is the licence check plugin and JOSM,
but as you say there is no rapid way to know exactly the status of an
object and of each one of its tags just by querying the API.

If we had the possibility to see how the bot algorithms will change
the data to make it compliant, re-mapping will be so much easier and
some experienced human contributors could do it even in regions where
usually they do not go and repair broken data on the fly.

Spotting errors just by looking around without even knowing if the
redaction bot has actually changed data in that particular region will
be time consuming and therefor never be done.

It's not good to put ourself in a hurry.

-- 
Sincerely 
Hendrik Oesterlin - email hendrikmail2002 at yahoo.de