[Imports] canvec-to-osm 0.9.6 now available

[Frank Steggink]

Hi James,

James A. Treacy wrote:
> Frank,
> Some more issues to add to the list.
>
> There are cases where the same way exists in different data sets but
> with different uses. A good example is that islands in the waterbody
> data are often repeated in the wooded data (if the island happens to
> be wooded, which happens a lot). Importing one set after another leads
> to duplicated ways. This case is time consuming to fix because you
> have to reconcile two ways and a relation. This same issue arises with
> wetlands. Probably others too but these are the ones I've encountered
> most often.
>   
Exactly, but it could be done automatically, provided that nobody else 
has made changes on the server already.
> It would be wonderful if this was handled automatically, but as you
> stated merging the different data sets for a more sophisticated
> analysis of the data would lead to massive files.
>   
Yes. I think that this is most problematic for the initial upload. When 
the data is downloaded afterwards, and changed (automatically or 
manually), the new changeset will be much smaller. The original nodes 
don't need to be reuploaded. I think they will be manageable, but we 
need to check how this really turns out.
> Another issue is how the data is partitioned. It would be really
> helpful if they were divided by geographical area. Currently there
> are a number of files sparsely populated with features (040P07 has 23
> files for Wooded_area). If you are working on one area you need to
> import all the files which defeats the purpose of splitting them in
> the first place.
>   
This actually means that the NTS tiles are too big to handle all at 
once. Maybe we should decide to use smaller working areas, by splitting 
up the NTS tiles in 4 x 4 subtiles. A disadvantage is that features will 
have to be split up (if this can be done automatically), or they have to 
be duplicated in the subtiles.
> Please don't suggest importing the current files one at a time as you
> can't download that large an area at once from openstreetmap. One
> could do a mass upload of each file but I thought that was frowned
> upon. Additionally, it means uploading without running the validator
> (resulting in duplicated nodes & ways).
>   
True, there is also the 50k nodes limit when downloading data. I'm 
already running into this when downloading the Quebec City area 
(021L14), and I haven't uploaded any single Canvec file there yet!
Eventually, all Canvec data will be uploaded one day, so this problem 
can't be avoided. (Unless OSM removes the 50k node limit.)

Frank