I'm not sure why you have asked for this to be unlocked while we are discussing this in a PM you initiated with me. Your test routes start and end within the DPM segments so you can't reliable be testing the DPM mechanism.

Please test routes that are longer, such as the test route I provide within my PM reply.

As stated by several Champs already, the name change of the ramp segment will not affect DPM as it doesn't match partial names, it only matches full strings. We have fixed many junctions which failed DPM when they included identical street fields through the junction but the full string didn't match;

• M40 (W), none, Oxfordshire to M40 (W), none, Warwickshire - failed DPM
• M6 (N), none, Walsall to M6 (N), Wednesbury, Walsall - failed DPM

DPM, like most mapping and mechanisms in routing, is a time based penalty added to routes, this does not absolutely prevent Waze suggesting the less prefered route, only heavily penalises it. You should be looking at factors that may actually change the behaviour at this junction, such as average speeds, which I suggested in my first PM reply.

Looking at route speeds along the direct route, the cut through segment is always showing average speeds of 14-15mph, this may be due to the segment being corrupted, or the current mapping including traffic waiting at the traffic lights skewing the data. Remapping the direct route, to clear any corruption, or deliberately splitting it close to the roundabout segment might improve things.

The route tester was setup on this junction at 16:30 on 20th Feb and has been checking the optimum route through the junction every 5 minutes since then. Up to 09:00 on 24th Feb it was recording a 9-10% failure in using the dedicated cut-through route, recording journey times at least 10 seconds faster than the expect cut-through.

Since then the route tester has not recorded a single failure, every suggested route has used the cut-through, even when predicted journey times spiked. No edits have been made in this time, so it appears that something in DPM has fixed itself.
Just a final note to remember that DPM is not a blanket ban on a route choice, like all routing choices it is simply a large time penalty that can be overcome in some conditions.

The requested change will make no difference whatsoever to detour prevention.

Changing from one non-continuous name to a different one will have no effect - other than to move from a country-standard "A40 > Gloucester" to a non-standard "To Gloucester and Ross".

In addition to being non-standard, the proposed new name would lead to the app saying "take the n'th exit on to to Gloucester and Ross"

Incidences of routing server issues are entirely common. I've heard a couple of complaints recently about unexpected off/ons that are caused by DPM not being correctly applied recently. Another one that's happened a few times recently is incidences of routing through private segments.

Some, at least, are probably caused by loss of connection to the routing servers, which can lead to the app using cached data for routing but with a far less sophisticated algorithm. But some are definitely routing server problems.

Within the last week, a major incident across the whole world, where Waze was giving wildly optimistic ETAs (average 50mph+ speeds across city routes) was the subject of an SOS thread with HQ. It is disappointing when our good work at mapping things correctly gets subverted by infrastructure problems beyond out control.

My fingerprints are slightly on this one as well

viewtopic.php?f=375&t=269659

I noticed on a drive through that Kate was saying "exit onto to...." so requested the change. Over 2 years ago...

It will be due to DPM being poorly implemented on the routing servers.

Is the implementation something that varies? E.g. I noticed 2 or 3 on-off URs in the past week, the first ones in a long time. Is this just a coincidence or would this have been due to the routing servers having "issues" at that particular time?

DPM may not be working, but it's not due to the fact that the junction is mapped wrong. It will be due to DPM being poorly implemented on the routing servers.

You're incorrect saying that DPM doesn't work if the road name partially matches the main route's name. The last segment of the possible detour before the 2nd bracketing segment must be different, yes, but it doesn't matter if it contains the road number as it does in this case.

Almost all motorway junctions in the UK are mapped with "Mxx > Somewhere" on the motorway entry segments. If this was the case, we'd be inundated with URs complaining about off/on instructions. Of all the instances of reports to Centercode of off/on issues, no one, not even Waze staff have said that the issues are being caused by having the road number as part of the name on the ramp segment.

In my opinion, DPM is properly mapped at this junction.

a2bpdq wrote:Anyone have an automatic tester I could use?

I have an automated tester that is now collecting data for the route through this junction.
It will be a week before we have a full picture of how it's performing.

a2bpdq wrote:A lot of details needed to be adjusted for the DPM to work. The final adjustment being the renaming of the ramp segment from "A40 (W)" to "To Gloucester and Ross". This change happened on 5th Feb 2018 and once that change went live in the app, all subsequent tests showed that DPM was working. Before that change, DPM was not working.

So I believe that having "A40" in the ramp name again is now causing DPM to fail.

The problem with your belief here is that you've misinterpreted the earlier change of name from "A40 (W)" to something not containing "A40" as a sign that DPM is doing partial string matching on road numbers, and that having "A40" somewhere in the segment name is therefore sufficient to break DPM here.

One of the four criteria that MUST be met in order for an alternate route to be classified as a detour is that the name of the segment must NOT match the name of the following segment on the primary route through the junction.

i.e. this segment MUST NOT have the same name as this one.

But note that this is nothing more than a simple comparison of the two name strings - there's no attempt to parse the string to extract any specific elements such as the road number. So it's no surprise that, in order to get detour prevention working last time, the ramp segment had to change its name away from "A40 (W)" given that this is the exact same name as on the following primary route segment. However, changing its name to anything else would have done the job, and still will provided the routing servers are working properly.