[Imports] Buildings & Address in Washington, DC, USA.

[Alex Barth]

Jason -


On Mon, May 26, 2014 at 6:35 PM, Jason Remillard <remillard.jason at gmail.com>
 wrote:

> Hi David
>
> Given that this is using the same process as the NYC import, what
> lessons/changes are planned to avoid the same problems encountered in NYC.
> I think NYC had to be paused a couple of times, it was not perfect. It
> would be dumb to do everything exactly the same.
>
>
This is a good question and I should sit down and write up the import
process and lessons of NYC better. Here's a quick brain dump:

- The technical approach w/ NYC is sound, I would do it again: make the
import repeatable, slice up tasks in pieces, offer those pieces in tasking
manager, upload manually, validate manually.
- Be prepared to stop and go. It's the nature of the beast. Based on the
complexity of an import there will be things that go wrong. You'll have to
be able to stop the import, go back and fix any issues you already
committed to the database. For the same reason it's incredibly important
that your conversion process is easily repeatable. You'll go back, fix and
rerun. Next time we'll be able to catch a lot more gotcha's I'm sure but
I'm also sure we won't catch all of them.
- Share everything in the open - all code and issues for the import are
here: https://github.com/osmlab/nycbuildings/
- Make ample time to engage within community, be prepared to slow down
where necessary to clear conversations as they come up. Our conversations
took more time than I thought they would and in hindsight I would make more
time for in person event or meetings.
- Make sure expectations of all involved are clear. This is something I
could have done a better job of. For instance, that we at Mapbox were going
to engage very directly in the upload process wasn't fully clear for
everyone. At the same time I think it was important for us to initiate this
import having the human resources to finish it up. There are already too
many imports that aren't finished.
- An import is actually not all that conducive to be a community building
project, especially not a large one. Imports are technically complex,
require experience with OpenStreetMap and commitment by those involved.
Community involvement is absolutely clutch, at the same time doing an
entire import is grinding work so people will loose interest over time.

At this point the NYC import is almost done in terms of data added. We are
now doing additional rounds of validation. As always it is open for
participation: http://tasks.openstreetmap.us/job/2
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/imports/attachments/20140605/d45dc75e/attachment.html>