[mkgmap-dev] [PATCH v2] Remove LineClipper artefacts
From WanMil wmgcnfg at web.de on Wed May 25 19:09:07 BST 2011
Am 25.05.2011 19:08, schrieb WanMil: >> Hi WanMil, >> >> On Tue, May 24, 2011 at 06:16:04PM +0200, WanMil wrote: >>> I cannot say for sure because I have to check that in detail but the >>> following problem might have happened. The autoclosing does not longer >>> prefer closing ways to connecting ways. >> >> Why not? What was wrong with autoclosing ways to connecting ways? > > Attached picture shows the problem. The mp gets two unconnected ways. It > must now decide if it is better to close the ways or to connect the > ways. Now I decided that it is better to use the shorter one. > >> >> Could the autoclosing issue some messages at the WARNING level, please? > > Do you really want to have warnings for that? You will get lots of > warnings... > >> >> By the way, I ran across to some Corine landuse imports that had split a >> polygon line because it contained so many nodes. I created a >> multipolygon out of that, with just the split lines in role=outer. I >> think that it is reasonable not to join connecting polygon lines, unless >> they are members in the same multipolygon. > > I didn't think of joining lines that do not belong to the same > multipolygon. > >> >>> So maybe the incomplete outer way was connected to an incomplete inner >>> way and therefore the way could not be closed and is therefore >>> abandoned. Maybe the algorithm should prefer connections of ways with >>> an identical role or better it should not connect ways with non >>> identical role. >>> >>> Sounds good. I will post a patch. >> >> I am looking forward to testing it. > > I implemented and tested this change with your usecase but it didn't > help. Maybe there is another problem... > > WanMil Ok, posted the patch which fixes your testcase. Please try. WanMil > >> >> Marko >
- Previous message: [mkgmap-dev] [PATCH v2] Remove LineClipper artefacts
- Next message: [mkgmap-dev] Splitter PBF support patch
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
More information about the mkgmap-dev mailing list