[Talk-us] [Imports] MassGIS Building Import Start

[Clifford Snow]

On Wed, Dec 12, 2012 at 7:15 PM, nicholas ingalls <
nicholas.ingalls at gmail.com> wrote:

> In regards to this and previous imports, I have found that this is by no
> means the law when it comes to imports. In Canada we have / and still are
> in the lengthy process of importing canvec data for the whole country. We
> do this in a distributed approach. Each user can upload the data with their
> own account. This is MUCH better in many cases as it encourages people to
> upload the canvec data in their own area instead of a user sitting hundreds
> of miles away uploading with a generic account. It also allows people to be
> held responsible (in a good way) for their data. With a generic account
> being used by multiple people mistakes are harder to track to the source.
> By using individual accounts problems can be more easily tracked. Ie a user
> is not fixing validator issues before uploading causing problems. With a
> single user account tracking down the specific uploader is near impossible.
>
>
I agree with this approach, using individual user id instead of having to
create a new import id. We have at least two types of imports, manually by
users and bulk done by scripts as well as bots that correct problems. For
manually imports I feel using the existing id should be sufficient. For
bulk imports, which large needs to be defined, a separate id might be
useful.

To make it real clear, to me a manual import is cutting and pasting one or
more elements (nodes, ways, polygons and multipolygons.) Bulk imports use
scripts, such as bulk_upload.py.

Where the Mass import is being done manually, I'd go with their plan to use
existing user ids. If the user screws up one import, the others might be
perfect. Would we expect to remove all of their other imports for one
error, or just expect the user to correct the problem? I don't think so.

It might help the community to explain exactly why a new id is necessary
and what harm not using a separate id creates. I know this criteria has
been discussed at length on the discussion list, but I have yet to register
any compelling reasoning.
-- 
Clifford

OpenStreetMap: Maps with a human touch
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk-us/attachments/20121212/5bb91cef/attachment.html>