AJ, Thanks for your message and your interest in this topic.
I have been reluctant to remove the natural=water tag from the relation (I am not the one that added it). I was worried that it might have widespread unintended consequences. However, given your and maxerickson's suggestion, I say we go ahead and do it. The rendering is very unpredictable the way things stand. The rendering may be fine early in the day, and then later, with no change to geometry or tagging, it is broken (e.g. islands "flooded"). "Dirtying" the tiles fixes the problem...until the next day. I also suggest we document this on the wiki, explicitly saying not to add the natural=water tag to the Great Lakes unless there is a discussion on these lists (thanks for including the Canadian list on your reply btw). Mike On Fri, Apr 24, 2015 at 9:23 AM, AJ Ashton <[email protected]> wrote: > On Sat, Apr 18, 2015 at 11:22 PM, Mike Thompson <[email protected]> > wrote: > >> User maxerickson sent me this comment directly about this issue: >> >> ===================== >> >> The current modeling of the Great Lakes is actually to use >> natural=coastline. >> >> The addition of natural=water to the lake superior relation is probably >> what caused the bad rendering at z13. >> >> If you check the history of the relation, you can see people repeatedly >> adding and removing natural=water. >> >> ===================== >> > > Yes, if Lake Superior is mapped as natural=coastline (which I think is the > easier-to-maintain approach for such a large & complex water body) then we > should remove natural=water from the multipolygon relation (r4039486). Does > anyone have any objection to this? It's causing some noticeable rendering > issues both in the standard style and for data consumers. > > There is also a second multipolygon relation for Lake Superior that > appears to be entirely redundant: > https://www.openstreetmap.org/relation/1120169 . It captures just the > Canadian half of the lake. I think this relation could just be removed > after going through it and confirming that all of its member ways are > properly tagged as natural=coastline (which they appear to be). Does anyone > have any reason to keep this relation? (cc'ing talk-ca) > > AJ > > _______________________________________________ > Talk-us mailing list > [email protected] > https://lists.openstreetmap.org/listinfo/talk-us > >
_______________________________________________ Talk-us mailing list [email protected] https://lists.openstreetmap.org/listinfo/talk-us
