On Thu, Apr 3, 2008 at 7:38 PM, Brett Henderson <[EMAIL PROTECTED]> wrote:
> I hope the job loss isn't too much of a downer, best of luck finding > something better. > > As for tiling, I hadn't considered polygons. They sound nasty. I'd > been thinking of something far simpler. For ways I was thinking of > splitting them at tile boundaries, adding synthetic nodes as required, > and creating new way ids where one way becomes multiple ways. Polygons > change all that. Initial thoughts are just detect closed ways and split > accordingly to make closed polygons inside each tile. It sounds like > that may not be appropriate though. Splitting polygon types according > to tags and rules adds a huge level of complexity and will constantly > require updates as new tags and polygon types are defined. I also get > the impression from your wiki page that this will be very Garmin > specific, and perhaps that's the only way to go. > > I'm going to have to back away slowly from this one and pretend I didn't > see anything ;-) > Well, I knew about my job loss since November. I have a couple prospects in the works (I'm an automation/controls engineer). Monday was my last day of work so this is my first few days of unemployment. I have a bit of severance pay, so it's okay so far. My proposed rules file on my Wiki page is currently targeted for making routable Garmin GPS maps, but I plan to make it generic enough that it could be used for tiling or filtering by tags, and possibly other uses. Ideally the same rules file could be used for tiling, filtering and then generating a routable map, by just picking the part that is of interest to the particular task. If OSM had a polygon type, or if every closed way was guaranteed to have a "area=yes" tag, then we could deal with tiling in a generic way, but since it doesn't (who knows why... I believe all professional GIS systems do, for a good reason), we have to add this big layer of complexity. Even if polygons were a separate type, there are still some tricky cases. A polygon would have to be split into multiple parts if the boundary slices off some "peninsulas". And what about multipolygons--if a "hole" intersects the boundary, then you would effectively remove the hole and then alter the path of the outer polygon to follow the boundary, then around the path of the hole, then back to the boundary again. I plan to get stuff working in phases--there's too much to get done in one shot. Karl
_______________________________________________ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/cgi-bin/mailman/listinfo/talk
