I've had a chance to play a little with routing today and noticed a couple of things. An example would explain this best so here goes. If routing from north of Manchester to somewhere along the M62 westbound then the route is as expected (M66 onto M62 at J18, M60). Bury to Roe Green is an example of this. Bury to Gorse Covert however starts on the M66, turns off at J3, goes through Whitefield and back on to the M60/62 at J17, travels past Roe Green and arrives at destination. Bury to CHester however starts along the M66, goes straight through junction 18 to the next junction where you go all the way round the roundabout, come back on yourself to junction 18 and then turn left onto the M60 where the route is then as expected.This odd behaviour then continues if routing to places further afield that require the same initial route. http://www.openstreetmap.org/?lat=53.5497&lon=-2.261&zoom=14&layers=B000FTF shows the junctions involved but I can't see how it can be data errors as some routes go M66, M62 westbound without problem and all 3 destinations mentioned above are on the same tile Additionally I will occasionally get the odd routes previously described when routing long distances but usually get the optimal route (this is from the exact same starting point. I realise there's a lot that's not yet fully understood in the img file so I just thought I'd report these observations in the hope that they can somehow add to the overall picture Cheers Paul P.S. The routes described are as given by my Edge 605 and not simulated through software
Hi Paul,
Bury to CHester however starts along the M66, goes straight through junction 18 to the next junction where you go all the way round the roundabout, come back on yourself to junction 18 and then turn left onto the M60 where the route is then as expected.This odd behaviour then continues if routing to places further afield that require the same initial route.
http://www.openstreetmap.org/?lat=53.5497&lon=-2.261&zoom=14&layers=B000FTF shows the junctions involved but I can't see how it can be data errors as some routes go M66, M62 westbound without problem and all 3 destinations mentioned above are on the same tile
I tried the Bury to Chester route on my eTrex and it doesn't go to the M66 but goes south directly to J17 on the M60. However, if I start my route a bit to the E of Bury on the A58 it does go down the M66 and then does the right thing and turns W around the Simister roundabout. So it appears to get that right. Can't explain why you're seeing badness there. Cheers, Mark
Mark Burton wrote:
Hi Paul,
Bury to CHester however starts along the M66, goes straight through junction 18 to the next junction where you go all the way round the roundabout, come back on yourself to junction 18 and then turn left onto the M60 where the route is then as expected.This odd behaviour then continues if routing to places further afield that require the same initial route.
http://www.openstreetmap.org/?lat=53.5497&lon=-2.261&zoom=14&layers=B000FTF shows the junctions involved but I can't see how it can be data errors as some routes go M66, M62 westbound without problem and all 3 destinations mentioned above are on the same tile
Can't explain why you're seeing badness there.
I might. If Paul used XAPI to extract a part of the country in order to build his Garmin map then it's possible that his XAPI excerpt isn't complete. I've been seeing more and more of this happening in South Wales where I'm doing exactly that. It seems (from other comments here) that there's a known problem with the various XAPI servers occasionally dropping a minute-diff and never recovering. I've seen this happening since the switch to the 0.6 API: I had never noticed it before that. I've been sent on bogus routes around Swansea (my home town) due to small bits of roads going missing. I've got a partially missing westbound carriageway on a dual carriageway in one particular case, I've also spotted a missing urban one-way street and a few other cases. If Paul looks carefully, does he spot (on the Garmin) a missing slip-road on a roundabout, responsible for there being no valid route from one side of a dual carriageway? Worse is an apparently present slip-road on a dual carriageway where the 'oneway' once pointed the wrong way, but has been fixed on OSM at some point, but where the fix was part of a minute-diff that the XAPI server dropped? Steve
If Paul used XAPI to extract a part of the country in order to build his Garmin map then it's possible that his XAPI excerpt isn't complete. I've been seeing more and more of this happening in South Wales where I'm doing exactly that.
It seems (from other comments here) that there's a known problem with the various XAPI servers occasionally dropping a minute-diff and never recovering. I've seen this happening since the switch to the 0.6 API: I had never noticed it before that.
I've been sent on bogus routes around Swansea (my home town) due to small bits of roads going missing. I've got a partially missing westbound carriageway on a dual carriageway in one particular case, I've also spotted a missing urban one-way street and a few other cases.
If Paul looks carefully, does he spot (on the Garmin) a missing slip-road on a roundabout, responsible for there being no valid route from one side of a dual carriageway? Worse is an apparently present slip-road on a dual carriageway where the 'oneway' once pointed the wrong way, but has been fixed on OSM at some point, but where the fix was part of a minute-diff that the XAPI server dropped?
I thought that XAPI was totally buggered at the moment. I haven't been able to use it for weeks (months?) I am currently using: http://download.geofabrik.de/osm/europe/great_britain.osm.bz2 Cheers, Mark
Steve Hosgood wrote:
Mark Burton wrote:
Hi Paul,
Bury to CHester however starts along the M66, goes straight through junction 18 to the next junction where you go all the way round the roundabout, come back on yourself to junction 18 and then turn left onto the M60 where the route is then as expected.This odd behaviour then continues if routing to places further afield that require the same initial route.
http://www.openstreetmap.org/?lat=53.5497&lon=-2.261&zoom=14&layers=B000FTF
shows the junctions involved but I can't see how it can be data errors as some routes go M66, M62 westbound without problem and all 3 destinations mentioned above are on the same tile
Can't explain why you're seeing badness there.
I might.
If Paul used XAPI to extract a part of the country in order to build his Garmin map then it's possible that his XAPI excerpt isn't complete. I've been seeing more and more of this happening in South Wales where I'm doing exactly that.
It seems (from other comments here) that there's a known problem with the various XAPI servers occasionally dropping a minute-diff and never recovering. I've seen this happening since the switch to the 0.6 API: I had never noticed it before that.
I've been sent on bogus routes around Swansea (my home town) due to small bits of roads going missing. I've got a partially missing westbound carriageway on a dual carriageway in one particular case, I've also spotted a missing urban one-way street and a few other cases.
If Paul looks carefully, does he spot (on the Garmin) a missing slip-road on a roundabout, responsible for there being no valid route from one side of a dual carriageway? Worse is an apparently present slip-road on a dual carriageway where the 'oneway' once pointed the wrong way, but has been fixed on OSM at some point, but where the fix was part of a minute-diff that the XAPI server dropped?
Steve
I use the Geofabrik mirror to download the great_britain.osm from http://download.geofabrik.de/osm/europe/great_britain.osm.bz2 I then use splitter r33 and usually the most current mkgmap to compile my maps using time java -Xmx2048M -jar -enableassertions mkgmap.jar --latin1 --gmapsupp --route --region-name="Great Britain" --region-abbr="GBR" --remove-short-arcs --stylefile=$stylefiles --max-jobs --net --code-page=1252 --country-name="GREAT BRITAIN" --country-abbr=GBR --tdbfile --tdb-v4 --description="Map of GB" --family-name="Great Britain" 700000??.osm $srtmdir/uk*.osm This also wouldn't explain why I could route from Bury to Roe Green turning west at J18 yet Bury to Chester goes straight through that juction only to turn round at the next and come back to J18 and then go past Roe Green Cheers Paul
participants (3)
-
Mark Burton -
Paul -
Steve Hosgood