Yes, I mean two non-contiguous private road segments.

It does appear that there are other issues at that destination.

vectorspace wrote:As I recall someone was marking vast areas of Ft Meade as private roads.

Well that was the guidance for private areas when the current properties of private roads were introduced.

vectorspace wrote: We have learned from txemt and Driving79 that this causes problems in residential areas, so it would on bases. I am considering changing the Private Installations guidance because of their work.

Seems to me that the private road type is intended for exactly this purpose. If it causes problems, seems to me we really need to raise that with the routing staff at waze.

The this I referring to is marking vast areas of Ft Meade as private roads.

Seems to me that the private road type is intended for exactly the purpose of marking private areas. If marking private areas with private roads causes problems, it still seems to me that we really need to identify such problems to the waze routing staff.

I'm in full agreement that we ought to understand how private roads function. I'm all for gathering all the information we have. But if we determine that private roads aren't functioning in the manner that waze has indicated they are intended to function, then I do think we should raise that issue with the programmers. I'm not saying we have to be paralyzed waiting for a response if there is an issue to be raised.

vectorspace wrote:Agree.... think we ought to fix the issue temporarily in the map and raise the issue to Waze.

txemt wrote:But is there an issue with private roads pertaining to military bases?

Agree with txemt, I'm not sure there is anything that needs fixing. I'll likely look to see if any issues arise with the private roads on the exits and then put them back on the entrances just to confirm that the timeouts don't return.

PhantomSoul wrote:Ok, let me see if I understand this correctly:

From what I know about Private Road (and Parking Lot Road works like this too, I think), there's a stiff penalty assessed to any route through a junction that involves going from any of the "public" road types to either Private Road (PR) or Parking Lot Road (PLR), AND vice versa. This means if you apply the single-PR per gateway checkpoint method to a base that has a main gated area and 1 inner gated area, you will incur this "transition penalty" between PR and the public road types 4 times - twice at the outer gateway when you enter and exit the PR segment, and twice again at the inner gateway when you enter and exit that PR segment.

Again agee with txemt. My understanding is that the transition penalty is only assessed once going from the PR or PLR to another road type.

PhantomSoul wrote:Now, is it possible when we test this for origins that are a relatively shorter distance from the base, it works fine, but if we try to route to it from, say 100 or more miles away, the combined 4 PR/PLR transition penalties, along with any other routing "scores" figured from calculating a route, cause an overflow situation? It sounds to me, based on the way PR/PLR is intended to be used, that Waze does not ever expect to encounter the transition between those two types more than twice: once if your origin is in a PR/PLR network to get you to public roads, and a second time if your destination is closer to a PR/PLR than any public road.

I really think waze only expects to ever encounter one transition if your origin is in PR or PLR network.

PhantomSoul wrote:Finally, is it possible that LiveMap or WME have bigger overflow thresholds than the Waze Client, which is optimized for the limited hardware and energy consumption of a smartphone? This would mean that certain routes calculated successfully on LiveMap or WME might time out in the Waze Client.

You're getting beyond my knowledge, but I was under the impression that when possible the route is calculated on the server side and sent to the client.

vectorspace wrote:I might not be understanding the logic here. You seem to be implying that along a path there is some maximum penalty upon which Waze aborts routing. Perhaps this is true, but I have not encountered this in the past.

I am pretty sure that there is a maximum something upon which Waze aborts routing. But there are several limitations on routes that are not necessarily penalty based, such as the limitation on routes that are too long as compared to the straight line distance between the origin and destination. So I'm not sure what all beyond just routing penalties contributes to an aborted route. However, this didn't seem to be an aborted route. When I ask for an overly circuitous route or even a route from Philadelphia, PA to Williamsburg, VA I get a message "Could not find route." However, in this instance I got either a message "Routing server timeout" or "Routing service timed out. Please try again later." These time out messages seem to be caused by different issues than Waze aborting the route.

vectorspace wrote:My assumptions from these tests were that Waze will find a route if it exists no matter what the penalty, that the penalty of a private segment is fairly extreme (at least equivalent to hundreds of miles of regular road), and that the accumulation of penalty can go very high and still route without giving up.

You do get to a point where waze will give up. For instance waze can't find a route from Washington, DC to Ocean City, MD if I have avoid toll roads on.

So the routing at Ft. Meade with the private roads on the exits rather than the entrances seems to be working fine. I can route on and off the NSA campus and the remainder of Ft Meade and can not route through these areas as expected. I have no problems with long distance routes that begin in the third level security zone and travel long distances. For example, I can route from the close NSA parking lots to Camp Beauregard or MacDill AFB without issues. Seems to me that private roads on the exits works identically in practice to private roads on the entrances.

Given that the tile changed the gates at the same time it corrected the corrupt segment, there is no reason to conclude the private roads on the exits are an improvement. Thus I plan to switch the private roads back to the entrances to make sure that the routing issue does not return and to comply with current practice. If anyone would like me to leave Ft Meade in its current state for a longer period of time for any further testing, let me know.

Can you clarify that the issue is with needing to cross two non-contiguous private segments? I think that's what you mean, just want to be sure.