> > In any case, I'm not sure that JTS uses FP equals for > anything other > > than checking whether two points are *exactly* the same. So this > > should be portable, I assume.
I think you're misreading what I said. I meant that JTS uses == (via equals2D) to test whether the *value* of two coordinates is exactly the same. Pointers don't enter into it. You absolutely should NOT try to replace those equality tests with "approximately equals". Approx equals may have a place in some of the algorithms, but this needs to be carefully evaluated on a case-by-case basis. Martin Davis, Senior Technical Architect Vivid Solutions Inc. www.vividsolutions.com Suite #1A-2328 Government Street Victoria, B.C. V8T 5G5 Phone: (250) 385 6040 - Local 308 Fax: (250) 385 6046 > -----Original Message----- > From: [EMAIL PROTECTED] > [mailto:[EMAIL PROTECTED] On Behalf Of > [EMAIL PROTECTED] > Sent: June 14, 2006 11:36 AM > To: GEOS Development List > Subject: Re: [geos-devel] QD (Quad-Double) Approach to Robustness > > > On Wed, Jun 14, 2006 at 11:16:09AM -0700, Martin Davis wrote: > > > Ok, but equality is actually used in JTS too. > > > Does Java have an automatic approximate equality ? > > > > I've been working under the assumption that Java uses a single FP > > model, so that it's safe to compare FP numbers. > > > > In any case, I'm not sure that JTS uses FP equals for > anything other > > than checking whether two points are *exactly* the same. So this > > should be portable, I assume. > > As far as you mean memory address comparison that would be > portable (pointers are integer values). Unfortunately we > don't have garbage collection for free in the C++ world, so > in some places your identity checks might have become value > comparisons. > > BTW, a quick grep on JTS sources for equals2D shows this is not true: > > ./algorithm/CGAlgorithms.java:197: } while > (ring[iPrev].equals2D(hiPt) && iPrev != hiIndex); > ./algorithm/CGAlgorithms.java:203: } while > (ring[iNext].equals2D(hiPt) && iNext != hiIndex); > ./algorithm/CGAlgorithms.java:214: if > (prev.equals2D(hiPt) || next.equals2D(hiPt) || prev.equals2D(next)) > ./algorithm/LineIntersector.java:284: if > (intPt[i].equals2D(pt)) { > ./algorithm/LineIntersector.java:311: if (! ( > intPt[i].equals2D(inputLines[inputLineIndex][0]) > ./algorithm/LineIntersector.java:312: || > intPt[i].equals2D(inputLines[inputLineIndex][1]) )) { > ./geom/Coordinate.java:134: public boolean > equals2D(Coordinate other) { > ./geom/Coordinate.java:159: return equals2D((Coordinate) other); > ./geom/CoordinateList.java:141: if > (last.equals2D(coord)) return; > ./geom/LineString.java:146: return > getCoordinateN(0).equals2D(getCoordinateN(getNumPoints() - 1)); > ./geomgraph/Edge.java:193: if (intPt.equals2D(nextPt)) { > ./geomgraph/Edge.java:234: if (! pts[i].equals2D(e.pts[i])) { > ./geomgraph/Edge.java:237: if (! > pts[i].equals2D(e.pts[--iRev])) { > ./geomgraph/Edge.java:253: if (! pts[i].equals2D(e.pts[i])) { > ./geomgraph/EdgeIntersectionList.java:145: boolean > useIntPt1 = ei1.dist > 0.0 || ! ei1.coord.equals2D(lastSegStartPt); > ./noding/SegmentNode.java:57: isInterior = ! > coord.equals2D(segString.getCoordinate(segmentIndex)); > ./noding/SegmentNode.java:81: if > (coord.equals2D(other.coord)) return 0; > ./noding/SegmentNodeList.java:69: > Assert.isTrue(ei.coord.equals2D(intPt), "Found equal nodes > with different coordinates"); > ./noding/SegmentNodeList.java:70:// if (! > ei.coord.equals2D(intPt)) > ./noding/SegmentNodeList.java:126: if (p0.equals2D(p2)) { > ./noding/SegmentNodeList.java:159: if (! > ei0.coord.equals2D(ei1.coord)) return false; > ./noding/SegmentNodeList.java:212: if (! > pt0.equals2D(edgePts[0])) > ./noding/SegmentNodeList.java:218: if (! > ptn.equals2D(edgePts[edgePts.length - 1])) > ./noding/SegmentNodeList.java:238: boolean useIntPt1 = > ei1.isInterior() || ! ei1.coord.equals2D(lastSegStartPt); > ./noding/SegmentString.java:157: if > (intPt.equals2D(nextPt)) { > ./noding/SegmentPointComparator.java:30: if > (p0.equals2D(p1)) return 0; > > --strk; > _______________________________________________ > geos-devel mailing list > geos-devel@geos.refractions.net > http://geos.refractions.net/mailman/listinfo/geos-devel > _______________________________________________ geos-devel mailing list geos-devel@geos.refractions.net http://geos.refractions.net/mailman/listinfo/geos-devel
