Is it right or left, Waze only gives exit on the typical side for each country. Perhaps the script needs to evaluate which country your in.Olestas wrote:Hmm, actually this was long time ago, just forgot to tell:
Shows as EXIT, but waze gives KEEP.
https://www.waze.com/editor/?env=row&lo ... ,153859370
As far as I know Livemap never says "Exit", it always says keep. The client app says exit when it is the proper road types, and it is on the typical side (the traffic side) of the road.FZ69617 wrote:Indeed, Livemap shows "keep right" for this route:Olestas wrote:Hmm, actually this was long time ago, just forgot to tell:
Shows as EXIT, but waze gives KEEP.
https://www.waze.com/editor/?env=row&lo ... ,153859370
However, Waze client app gives an "exit right" instruction here:
Exactly what JAI suggests in this particular case.
On the bright it will be much simpler again, and this time we know what to expect.Olestas wrote:Omg, not again...
I don't anticipate losing any instructions, we may get more where there currently is BC.taco909 wrote:Will this cause any issues with wayfinders that were added last summer when BC was made "more liberal" and we lost instructions where we had them before?PesachZ wrote:On the bright it will be much simpler again, and this time we know what to expect.Olestas wrote:Omg, not again...
No that is not a bug, BC doesn't prefer a lower or 0 angle at all, as long as it's less than 45, that's the only angle that matters.taco909 wrote:Ya, my concern was possibly doubled instructions.PesachZ wrote:I don't anticipate losing any instructions, we may get more where there currently is BC.
In rare cases you may lose an instruction, but it would be a very small subset of cases.
But I think I've seen the bug you are referring to... is that perhaps where the only match is the road type and it pushes BC through a "stay" movement rather than the straight-through that is a 0-degree move? We had a bad one up in Lancaster.
This isn't really related to the the script, and if there are specific cases we need to work on let's move the discussion to the BC change thread.taco909 wrote: California drivers are dumb and want an instruction when the road name changes even if they are going straight... imagine the confusion this created with no name matches, but BC was on the "Stay to the right" (which was really very close to being more appropriate as "turn")
Thanks for posting. This is line with the recently announced changes to the BC algorithm. The algorithm is being simplified so any name (primary or alternate) is given equal weight.voludu2 wrote:I have been very much enjoying the new turn instruction prediction in JAI. I found a case where JAI does not reach the same conclusion on BC as the mobile app or route checker:
https://www.waze.com/editor/?env=usa&lo ... 8,75487051
Waze announced "stay to the right" because Sin Name matches Sout name and also S1 alt name. JAI does not catch on to this.
Considering that a self connected U-turn (only one segment) is able to generate a prompt, it seems to reason they adjusted the algorithm even for two segments in the case of a U-turn prompt.FZ69617 wrote:Thanks for finding this another anomaly.Arnoniem wrote:I just noticed one of my tests showing a "turn left" at the 168.45 angle. It could be the bug that gives a left instead of uturn, but I don't think so, since the livemap gives the same result. Have a look here:
https://www.waze.com/editor/?env=row&lo ... ,274372422
Foreseeing other unknown cases the gray zone has been extended into 1 degree (i.e. 168.24 ± 0.5).
Hmm, another interesting change to the algorithm. Previously, no instruction was the only option in case of just one s-out.Arnoniem wrote:Oh, and i noticed another thing:
On a 2 segment uturn at 170degr+, JAI gives a Best continuation, however the actual instruction is also a uturn:
https://www.waze.com/editor/?env=row&lo ... =280755690
This JAI development version fixes the problems.
Thank you Arnoniem for the reports.
Re: [script] WME Junction Angle Info