On Mon, Nov 20, 2017 at 6:15 PM, OSM Volunteer stevea <stevea...@softworkers.com> wrote: > Please, ENTER data using shared ways where it makes sense to do so. Nobody > is saying "don't do that." ALSO, please be aware that existing > NON-multipolygon data (especially imports and other "curated" data) may very > well suffer from the process of being "multipolygonized." There is a balance > to be struck, and I would be very disheartened to see our map become "dumbed > down" by data which should be multipolygon somehow become twisted into not.
The cases where I have intentionally converted curated, imported data to shared ways have all been in imports that I curate myself. I wind up treating the result as being no different as any other imported object that's subsequently been modified by a local mapper. If a reimport discovers that the last user to touch the object wasn't the import user ID, it flags the upstream data for manual conflation, and does nothing. Even when it does find that it was the last modifier of the object, the most it does is to prime JOSM with the proposed change and let me confirm and commit. The largest data set that I work with has a couple of thousand polygons. I've done two reimports, each of which have changed a couple of dozen. Of those, a handful, few enough to count on fingers, have been subsequently modified by local mappers. While the initial import ran over a couple of months, an annual reimport takes me maybe a couple of evenings. Yes, I did hear a sentiment calling for "dumbing down" the map - and it wasn't so much you specifically, as that the early returns appeared to be shaping up into yet another Europe-North America divide. I know that your position is considerably more subtle; we've discussed this one fairly extensively already. I just really want to nip the idea in the bud that simplicity is preferable to correctness. I'm also fairly cross, because I've discovered in the last week that two large multipolygon relations that I had spent hours on setting up had been dismembered. Since they were born in an import (however much manual editing and conflation was done), I'm not comfortable with restoring the data, even though significant information has been lost by the manual changes. In both cases, the topology was damaged by the Potlatch user, who then 'repaired' things by moving tagging that had been on the relation to some, but not all, of the outer ways of the modified object. These were two different users. I suppose this reinforces both the case that multipolygons should be used only as a last resort and that imports damage the community, but I'm sad to see regions of the map lose tagging that had previously been informative, and begin to feel that my hands are tied from repairing them. _______________________________________________ Talk-us mailing list Talk-us@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk-us
