Hi all, I've commited r249 - overlap defaults to -1, it is set to 0 if keep-complete=true, else to 2000 - using a bounding polygon forces no-trim=false. Splitter makes sure that no holes are created within the polygon - if using a bounding polygon splitter will not create tiles that cover large parts outside of the polygon, still, tiles will be aligned and therefor may cover small outside areas - too complex polygons are rejected, esp. those that are not rectilinear The split algorithm is now quite complex, I am pretty sure that I did not find all errors yet, but it works fine for all workload that I have tested. The performance is poor if you split large areas with very few points, e.g. an ocean. Gerd -- View this message in context: http://gis.19327.n5.nabble.com/splitter-r249-tp5738211.html Sent from the Mkgmap Development mailing list archive at Nabble.com.
Am 29.11.2012 17:24, schrieb GerdP:
Hi all,
I've commited r249
- overlap defaults to -1, it is set to 0 if keep-complete=true, else to 2000 - using a bounding polygon forces no-trim=false. Splitter makes sure that no holes are created within the polygon - if using a bounding polygon splitter will not create tiles that cover large parts outside of the polygon, still, tiles will be aligned and therefor may cover small outside areas It works fine now. Thanks a lot. If splitter makes always sure that no holes were created, --no-trim would be obsolete I think.
Henning
Henning Scholland wrote
Am 29.11.2012 17:24, schrieb GerdP: It works fine now. Thanks a lot. If splitter makes always sure that no holes were created, --no-trim would be obsolete I think.
Fine :-) thanks for testing! reg. no-trim: it is difficult to say what splitter should do if you don't specify a bounding polygon. Do we only care about holes that are enclosed by other tiles or do we also not want gaps between two tiles? The latter typically happens when you have some islands . The tiles with islands are not trimmed, but tiles between them may be trimmed. I guess this is also not wanted? Gerd -- View this message in context: http://gis.19327.n5.nabble.com/splitter-r249-tp5738211p5738307.html Sent from the Mkgmap Development mailing list archive at Nabble.com.
Am 30.11.2012 00:47, schrieb GerdP:
reg. no-trim: it is difficult to say what splitter should do if you don't specify a bounding polygon. Do we only care about holes that are enclosed by other tiles or do we also not want gaps between two tiles? The latter typically happens when you have some islands . The tiles with islands are not trimmed, but tiles between them may be trimmed. I guess this is also not wanted? Hmm... it's hard to say. ;) Maps looks nicer in MapSource&Co with --no-trim, if you generate sea. On the other hand it could cause problems, because coastline-nodes aren't counted and so a tile could get to big for mkgmap. Eg. if you generate a map of Italy you may get hole coastline of Adria. So in this case it would be better not to extend the map like --no-trim did.
Henning
participants (2)
-
GerdP -
Henning Scholland