[Talk-ca] Imported Canvec buildings with multiple nodes

Frank Steggink steggink at steggink.org
Tue Jan 26 23:58:59 GMT 2010


Hi,

The duplicate building nodes can be (more or less) efficiently fixed 
with JOSM. When you install the Validator plugin (perhaps it's already 
done by default by now), you'll see which nodes are duplicated. The 
simplest way to solve them would be to use the Fix button.

However, generally it is better to be prudent, because a series of 
duplicate nodes indicates that a way is likely duplicated as well. Those 
cases need to be resolved as well. When using the Validator plugin, 
overlapping ways are also detected, but they're considered a mere 
warning. I believe that a way is only considered when the nodes are the 
same. So, you'll either get duplicate nodes or overlapping ways.

In JOSM you can also make a selection (graphically, or using the Search 
form), and when you use the Validator with a selection, only the 
selected elements are being validated. So, you can search for Canvec 
buildings only. NB: it is not possible using boolean operators (the 'OR' 
operator often seems to fail), but you can perform multiple searches in 
a row, and choose how the query should affect the selection (replace, 
add, remove, search within).

An example, to select Canvec buildings:
* Select 'building=yes' (without quotes), with 'replace selection' checked.
* Select "canvec:CODE" (with quotes!), with 'find in selection' checked.
This will select buildings imported which have a 'canvec:CODE' attribute.

Note that the current rules.txt file doesn't include a "buidling=yes" 
attribute for all buildings. When doing the import near Quebec City 
(tile 021L), I added that code manually. All cases I have evaluated are 
indeed buildings. And it wouldn't make sense to put non-buildings in the 
buildings layer. Maybe Sam can comment on this.

Speaking about duplicate nodes: I still need to use the Validator to 
solve them in the Charlevoix area. I imported the forested areas without 
merging them in existing OSM data, because there were no existing 
forests. However, there were already water features adjacent to those 
wooded areas. So, a note to people doing imports: please make sure that 
any validation errors are resolved as much as possible. Not all errors 
can be resolved (like some incorrect tag/value errors), but the 
graphical errors can. And re. warnings and other things found during 
validation: that is almost impossible. Maybe the validation rules of 
JOSM need some finetuning.

Cheers,

Frank




More information about the Talk-ca mailing list