Hi Peter, the result I posted is not the result of oneway-subnetwork procedure.
The total procedures include: 1. remove zero-degree node 2. findSubnetwork 3. oneway-subnetwork procedure 4. findSubnetwork again on graphhopper.cleanup() My question is, why those islands are recognized on step 4, but not on step 2? *Best Regards,* *ZhiQiang ZHAO* On Tue, Jul 28, 2015 at 12:02 AM, Peter <[email protected]> wrote: > Hi ZhiQiang, > > I think it is because both networks are oneway subnetworks not found by > the normal subnetwork procedure (but by the oneway-subnetwork procedure) > and you defined the oneway minimum size to 20 > > Regards, > Peter > > > On 28.07.2015 03:13, John Zhao wrote: > > Hi Peter, > > What I do is: > 1. minOnewayNetworkSize = 20, minNetworkSize = 200 > 2. build san francisco bay area osm data > 3. I print out the subnetworks result of the second call. > > int remainingSubnetworks = preparation.findSubnetworks().size(); > > 4. I found the subnetwork has some smaller than 200, like: > > subnetwork start from: 37.32611992939085,-121.9961998312816 size: 24 > > subnetwork start from: 37.78373608999855,-122.25065187925067 size: 34 > > 5. I can't understand why the subnetworks with 24 nodes and 34 nodes are not > removed by preparation.doWork(); > > It call the same method: > > Map map = this.findSubnetworks(); > > > *Best Regards,* > *ZhiQiang ZHAO* > > On Mon, Jul 27, 2015 at 12:54 PM, Peter <[email protected]> wrote: > >> Hi John, >> >> sorry, I do not understand your problem or question here. Would you >> describe it again step by step for me :) ? >> >> Kind Regards, >> Peter >> >> >> On 27.07.2015 21:45, John Zhao wrote: >> >> Hi Peter, >> >> Thanks. >> Actually I only have 1 flagEncoder in the EncodingManager. >> The call is exact same, preparation.findSubnetworks() >> >> preparation.findSubnetworks() using edgeFilter which is also from >> singleEncoder. >> >> >> *Best Regards,* >> *ZhiQiang ZHAO* >> >> On Sun, Jul 26, 2015 at 7:56 AM, Peter <[email protected]> wrote: >> >>> Hi John, >>> >>> it should not be related to calling these method twice. It is just one >>> time where you calculate the subnetworks independent of any FlagEncoder or >>> direction via findSubnetworks and the second pass is FlagEncoder- and >>> access-dependent via removeDeadEndUnvisitedNetworks. >>> >>> Regards, >>> Peter >>> >>> >>> On 24.07.2015 21:16, John Zhao wrote: >>> >>> Hi Peter, >>> >>> I am still confused. >>> at first we call >>> map = findSubnetworks(); >>> >>> after the cleanup, we call the same method in Graphhopper. >>> >>> int remainingSubnetworks = preparation.findSubnetworks().size(); >>> >>> Why the subnetwork was recognized the latter time, but not the first time? >>> >>> we remove some edges make it not connected? >>> >>> >>> *Best Regards,* >>> *ZhiQiang ZHAO* >>> >>> On Thu, Jul 23, 2015 at 2:22 PM, Peter <[email protected]> wrote: >>> >>>> >>>> Hi ZhiQiang, >>>> >>>> yes, according to the wiki this is wrongly mapped: >>>> * Avoid tagging highway intersections as that does not make clear which >>>> way has the impediment. * >>>> >>>> http://wiki.openstreetmap.org/wiki/Tag:barrier%3Dgate >>>> >>>> Peter >>>> >>>> >>>> On 23.07.2015 23:16, John Zhao wrote: >>>> >>>> Hi Peter, >>>> >>>> Maybe the following one related with >>>> https://github.com/graphhopper/graphhopper/issues/388#issuecomment-88066385 >>>> >>>> >>>> I have a look at 37.32611992939085,-121.9961998312816. >>>> It seesm related with barrier=gate at intersection. >>>> http://www.openstreetmap.org/node/1126492194 >>>> >>>> *Best Regards,* >>>> *ZhiQiang ZHAO* >>>> >>>> On Thu, Jul 23, 2015 at 2:11 PM, Peter <[email protected]> wrote: >>>> >>>>> >>>>> There are two types of subnetworks and the smaller ones seems to be >>>>> 'one-way subnetworks' which means they are eg. only reachable as >>>>> destination or start. But if you would start from a destination-only >>>>> subnetwork you'll get 'not found' for all points outside of this network. >>>>> >>>>> Regards, >>>>> Peter >>>>> >>>>> >>>>> On 23.07.2015 23:03, John Zhao wrote: >>>>> >>>>> Interesting, >>>>> when I increase minOnewayNetworkSize from 20 to 50, the following two >>>>> disappeared. >>>>> subnetwork start from: 37.32611992939085,-121.9961998312816 size: 24 >>>>> subnetwork start from: 37.78373608999855,-122.25065187925067 size: 34 >>>>> >>>>> *Best Regards,* >>>>> *ZhiQiang ZHAO* >>>>> >>>>> On Thu, Jul 23, 2015 at 1:55 PM, John Zhao <[email protected]> wrote: >>>>> >>>>>> Hi, >>>>>> >>>>>> I tried car flag encoder with following parameter on San Francisco >>>>>> bay area data from mapzen. >>>>>> >>>>>> https://s3.amazonaws.com/metro-extracts.mapzen.com/san-francisco-bay_california.osm.pbf >>>>>> >>>>>> minNetworkSize=200 >>>>>> minOnewayNetworkSize=20 >>>>>> >>>>>> I printed all the remaining subnetworks. >>>>>> edges: 591932, nodes 437420, there were 3496 subnetworks. removed >>>>>> them => 13121 less nodes. Remaining subnetworks:5 >>>>>> The remaining subnetworks are: >>>>>> subnetwork start from: 37.32611992939085,-121.9961998312816 size: 24 >>>>>> subnetwork start from: 37.56018439442332,-122.30257814308803 size: >>>>>> 436637 >>>>>> subnetwork start from: 37.78373608999855,-122.25065187925067 size: 34 >>>>>> subnetwork start from: 38.180185962770565,-121.70631393878864 size: >>>>>> 301 >>>>>> subnetwork start from: 37.85717050411933,-122.07633641532816 size: 424 >>>>>> >>>>>> I don't understand why there is still subnetwork less than 200 >>>>>> nodes. >>>>>> >>>>>> I have a look at 37.32611992939085,-121.9961998312816. >>>>>> It seesm related with barrier=gate at intersection. >>>>>> http://www.openstreetmap.org/node/1126492194 >>>>>> >>>>>> *Best Regards,* >>>>>> *ZhiQiang ZHAO* >>>>>> >>>>> >>>>> >>> >>> > > _______________________________________________ > GraphHopper mailing list > [email protected] > https://lists.openstreetmap.org/listinfo/graphhopper > >
_______________________________________________ GraphHopper mailing list [email protected] https://lists.openstreetmap.org/listinfo/graphhopper
