[Imports] Making the current guidelines/code of conduct about imports/automated|mechanical edits clearer and merged
Glen Barnes
barnaclebarnes at me.com
Wed Dec 19 00:31:25 UTC 2012
On 19/12/2012, at 12:43 PM, Clifford Snow <clifford at snowandsnow.us> wrote:
> Secondly, the need to use a new user id for imports. Looking back in history on the imports mailing list, it seemed at one time that the requirement for a unique user id for imports was just for mechanical bots and scripted imports. I can understand the thinking. But somewhere along the way, that requirement seemed to be for all types of imports. Rather than ~100 objects, I would recommend separating mechanical bots and scripted imports from manually cutting and pasting.
Agreed. If you are doing a 'managed import' of bulk data (rather than an automated import) then you may have 100 volunteers. Do they each need a separate upload account? Should they all share 1 login?
> It was the ~100 objects that first caught my eye. What is an object. A node or a way. I might import a stream. it could easily have more than 100 nodes yet be fairly trivial to import. But 100 streams is a lot of objects to be manually importing at a single shot.
It is hard to know what is right in this instance. For the linz2osm tool we've limited checkouts to 100 features (a feature being a node for point sets and a way for polygon sets). This generally works well but sometimes you need to checkout less and sometimes you wish you could checkout more. How many is the right amount comes down to how comfortable you feel when managing the dataset in JOSM and how confident that you didn't miss anything. The number can also change within a dataset. Checking our 100 islands in an undermapped area is easy but doing the same in a highly mapped region can be a lot more work in checking.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/imports/attachments/20121219/ce25a9ac/attachment-0001.html>
More information about the Imports
mailing list