Am 13.07.2012 10:13, schrieb Paul Norman:
From: nimix [mailto:[email protected]] Sent: Friday, July 13, 2012 12:58 AM To: [email protected] Subject: [OSM-dev] (Multi)Polygon handlingFrederik Ramm wroteIf we can't get everyone to use the same code base, then it would at least be great to reach some kind of agreement here, or maybe at least produce a test suite that everyone runs their code against.I created a testsuite for that with some more cases that need special treatment. Unfortunately I only stored the WKB representation in Google mercator projection, so it is not very human readable. The testcase contains a LINESTRING and a MULTIPOLYGON for simple closed ways and for relations a MULTILINESTRING and a MULTIPOLYGON ([1]). I would volunteer to bring it in an other format so that it is useful for everyone writing converters and I would like to discuss what is the best way to do it. I would suggest an .osm file and a Wiki Page containing the expected results in WKT/WKB and some explanation for each case. Any other ideas? Best regards, MelchiorMy following of web standards has convinced me it is necessary to define the expected output in every case, including invalid ones, or you will have differing implementations. I've been giving it some thought as to what needs to be tested, and I don't think it's excessive. I'd estimate 25-50 small tests would cover everything.
+1 But it should be possible to reject false polygon models. So to say:- any application "must" deal with correct tagging in the same way (which is defined by the corresponding test cases) - if an application interprets wrong polygons, it "must" do it in the way which is defined by the corresponding test cases - but(!) every application is allowed to reject these polygons because of errors - any application is encouraged to report wrong polygons detected to users who may be able to fix that in the database.
("must" of course only to correctly claim "standard" osm polygon handling)
regards
Peter
_______________________________________________
dev mailing list
[email protected]
http://lists.openstreetmap.org/listinfo/dev
