[OSM-talk] Duplicate nodes incorrectly removed by bot BugBuster ?

Ruben Wisniewski ruben at all-in-si.de
Sun Sep 27 16:28:19 BST 2009


Hi there,

just want to join the discussion, this is my first post to this list,
and till now I don't followed the list.

BugBuster is my Bot.

woodpeck note me that you discuss it here.

Pieren schrieb:

> > Yes, it seems it is only touching buildings, maybe created by a
> > previous import because they contain a tag "ID".
> >   
>   
No I'm not only touching this buildings, I tried to fix every doubled
node. It's the same thing a user would do with a josm validator.

It might look like just editing this because the bot has the sorted ways
from the planet file as source and the import seems to be very fast so
the IDs of the ways are very close together.

Eugene Alvin Villar schrieb:

> > Nope. That way is not closed. A closed way is a way has the *same* starting
> > and ending nodes, not a starting and an ending node that happen to have
> > identical coordinates. So the way is still broken, but the bot's fix
> > (duplicated nodes) is correct for the particular bug the bot is fixing.
>   
You're right the bot only fix this one bug in this run. I doesn't fix
the broken import script that has created the buildings before. Just
because I didn't know about it.


Shaun McDonald schrieb:

> > Another bad thing about that bot is that it is producing a changeset
> > for each edit, with nearly 10 changesets per minute.
> >
>   
I'm sorry, didn't thought this is a problem, I just heard before that
often somebody dislike that changesets changed things over the whole
world and ordering the ways which are near together in location is much
work if I just want to use the planet file as datasource again. I might
change to exports of parts of the world and upload changsets for the
parts. If you (all) like this way more. Or I might merge the changesets
to big changesets from the IDs from the ways which leads in changesets
with boxing around the world.


Greetings Ruben






More information about the talk mailing list