rinigus | # 951 | 2020-01-03, 16:25 | Report

I don't want to enable it by default, but it can be something users can switch on

Kabouik | # 952 | 2020-01-03, 18:16 | Report

Of course, it should not be forced Catbus.

Does Valhalla support waypoints Rinigus? This could be a way to force alternate routes, and would be convenient anyway.

rinigus | # 953 | 2020-01-03, 19:01 | Report

Waypoints are supported by Valhalla. But I will need to rewrite navigation part of Pure Maps to get it working. That's planned.

pichlo | # 954 | 2020-01-07, 09:10 | Report

I am sorry, rinigus I used Pure Maps again after a long gap, to get me to an address about 25 minutes away*) and guess what? The perennial issue with the hang background process after closing the UI popped its ugly head again This was on an Xperia X, with no other foreground processes running, so I cannot use an excuse of low resources.

____
*) Pure Maps said 12 minutes, Bing said 21. Bing was much closer to reality. It was by no means the first time that Pure Maps severely underestimated the journey duration, but that is a topic for a whole other discussion.

rinigus | # 955 | 2020-01-07, 09:40 | Report

Ohhh, no! And I though all is fine with Pure Maps and I can work a bit on other projects before coming back to it. (which I am still going to do though). Although, I still don't know what goes wrong with this background hang. But we can try to debug it again in the future.

Re estimated time: don't shoot the messenger. Pure Maps is not estimating time itself, but the router is. Where I live, Valhalla via OSM Scout Server estimates time quite well. So, maybe UK times have to be somehow adjusted in Valhalla's side - if its the one you use.

nthn | # 956 | 2020-01-07, 09:42 | Report

Perhaps Bing takes into account (live) traffic information, or maybe there's a wrong speed limit set in OpenStreetMap for the roads you were driving on?

pichlo | # 957 | 2020-01-07, 10:03 | Report

Services like Bing or (cough!) Google sure do. They even give you alternatives.
But I noticed the overly optimistic Pure Maps estimates even in situations with no traffic at all. Like on early Sunday morning.

It is usually OK-ish on longer journeys (more than say 2 hours). I have learned to double the estimate on journeys under 30 minutes.

Don't worry, I do realize it is OSM's fault

An interesting point about the UK specifics. The Pure Maps' estimate was 12m 44s, Bing's 21m. The ratio is 1.6, suspiciously like km vs. mile. I may need to look into my settings.

taixzo | # 958 | 2020-01-07, 14:51 | Report

I suspect the short vs long discrepancy may be because Pure Maps is not including time spent at stop lights, stop signs etc. On longer trips these make up a smaller fraction of the total because you're more likely to spend more time on highways.

rinigus | # 959 | 2020-01-07, 20:42 | Report

Let's not mix the messenger (Pure Maps) and routing engine (Valhalla), unless I managed to screw up time conversion for UK.

Valhalla does take into account time spent on traffic lights, left, right turns (should be in accordance with the country). Question is though how accurate the corresponding delays are, how good data is available at OSM.

pichlo | # 960 | 2020-01-08, 09:23 | Report

A few results to compare the estimated journey times. In all cases the source and destination were the same and all three engines planned the same route. Pure Maps profile Offline, maps OSM Scout (default), units British.

Route 1, 12 miles (home to a random nearby location)
Pure Maps: 15 min
Bing: 25 min
Google: 22 min
Reality: around 25 minutes IIRC, heavy but flowing traffic, no delays

Route 2, 34 miles (home to work)
Pure Maps: 37 min
Bing: 48 min
Google: 44 min
Reality: about 45 minutes, light traffic

Route 3, 98 miles (home to our most commonly used airport)
Pure Maps: 1 h 31 min
Bing: 1 h 46 min
Google: 1 h 45 min
Reality: usually between 1 h 30 min and 2 hrs, mostly on motorways, exceeding speed limits (shhh)

OSM Server has its own units setting. This was set to Metric. Changing that to Imperial and running the above tests again showed no difference, so there goes my theory. It looks like taixzo hit the mark.