Re: [Announce] Poor Maps
 

Re: [Announce] Poor Maps
 

Valhalla's team is joining Mapbox, see announcement at https://github.com/valhalla

Re: [Announce] Poor Maps
 

Interesting. Good for development and offline use, I guess, but quite impossible for online service use. Mapbox's terms and pricing are very difficult, e.g. "You may only use our Navigation SDK or results from our Directions API in or for real-time turn-by-turn navigation (regardless of mode of transportation) for up to 50 Seats."

Re: [Announce] Poor Maps
 

So, navigating a double-decker bus is strictly prohibited? I can stand it :)

Re: [Announce] Poor Maps
 

Well, we can always setup a service for Valhalla online. I have to make imports for the planet anyway, so we have data available for routing. Same goes for Mapbox tiles, btw. I don't know what are the requirements for the server, but its surely possible if the users would pay for it. Or if some company interested in it would like to back it. Let's see what is going to happen and then look for solutions.

But loosing a company such as Mapzen is a loss. In addition to Valhalla, they were supporting the developer of libpostal.

Re: [Announce] Poor Maps
 

Well, here's a summary of the routing providers off the top of my head.

• MapQuest Open: probably a deprecated service after MapQuest's revamp a while back: "They are currently still maintained but their future is still being determined. Sorry, not much more news than that at this time."
• OSRM: Demo server is very unstable, not usable in production even for a small user base.
• Mapbox: Impossible terms unless paying $500/mo.
• HERE: Impossible terms.
• GraphHopper: The last I looked, they had good routing, but really bad turn-by-turn directions. Will need to recheck this.

So, there might be need to set up a server, but no hurry while MapQuest Open works. And, of course for the current user base who install their own phone OS, running OSM Scout isn't unreasonable either, even though I prefer the ease of online.

Re: [Announce] Poor Maps
 

Testing on Saera (which still has an optional GraphHopper backend) the directions are still awful - the directions on the route I tested were basically "Head north", "Continue", "Continue", "Continue", "Continue", "Turn left on X street" - where the Continue's should have been interpreted as "Take the ramp onto the interstate", "Merge onto the interstate", "Take exit Y", and "Merge onto Z street".

Edit: Also might want to add Google (impossible terms) and Bing (impossible terms) to that list.

Re: [Announce] Poor Maps
 

does openrouteservice.org provide an API or is it already listed here?

Re: [Announce] Poor Maps
 

I hadn't noticed, seems like they might have turn-by-turn directions, even though they are not displayed on their web demo. I'll look into it.

Re: [Announce] Poor Maps
 

Thanks for that, was looking for a free geocoding service and you've just linked one :D

Re: [Announce] Poor Maps
 

There is turn-by-turn, its just hidden in a pull-down of the side menu

Re: [Announce] Poor Maps
 

According to https://github.com/GIScience/openrouteservice, its using "forked and edited version of graphhopper 0.9". Its probably similar to what @taixzo tried earlier.

Re: [Announce] Poor Maps
 

Found it now in the web demo. Looks the same as I remember: full of continues and slight turns when there's no actual maneuver and not quite descriptive of the actual maneuvers.

Re: [Announce] Poor Maps
 

Today I finally had the opportunity to test navigation with PoorMaps (version 0.34) and OSM Scout Server.

It worked very well for me. :)

A few issues, which might be improved:

• Using tiles "OSM Scout cat night @2x" you get a blue map, with a blue route/track and a blue position marker (where am I). This is very difficult to "read". Maybe the colors for route and the position marker could be changed, to work well on "day" tiles and "night" tiles. Or give the user ability to change those colors.
• Voice navigation in German works. But the voice is using the English term "meters" (as in "turn left in 300 meters"). In German using "Meter" (without "s") would be correct.

Should I file GitHub issues for these points?

Thanks for your good work! :)

Re: [Announce] Poor Maps
 

No need. Night icons is a known issue, likely to be fixed eventually. The translation issue just means that the German translation is incomplete, anyone can fix that. Translations are a bit difficult since it's a mixture of translations from a router API and translations shipped with Poor Maps, so the chance of them being "out of sync" is somewhat unavoidable.

Re: [Announce] Poor Maps
 

Hi, can you add google maps to choose them in application, google maps have newest data base of streets and all changes. I use navigation very often but I have SailfishOS on Xiaomi Redmi Note 4 so I can`t use android apps.

Re: [Announce] Poor Maps
 

Looking at https://www.transifex.com/otsaloma/poor-maps/, I see "German (Germany) 100%". So I think this is not an issue, which could be fixed via transifex.

I found two posts in this thread dealing with the "meter(s)" problem already:

• https://talk.maemo.org/showpost.php?...4&postcount=97
• https://talk.maemo.org/showpost.php?...4&postcount=98

If I understand correct, the distance unit is no translateable object yet.

Re: [Announce] Poor Maps
 

No, it's against Google's terms.

It is translatable. But, sorry, I didn't check, it is already translated to German in Transifex, but that has been after the last release. So, the issue will be fixed with the next release, which is likely to happen around first of February.

Re: [Announce] Poor Maps
 

good joke. Google maps is in my area by far the most outdated database I have checked in a while.

Re: [Announce] Poor Maps
 

Coverage varies, but where Google Maps is good, it's really good, way ahead of the competition.

https://www.justinobeirne.com/google-maps-moat

Re: [Announce] Poor Maps
 

The competition is more than just Apple Maps. And counter-example, here in the Netherlands OpenStreetMap is often much more detailed than Google Maps, and much more up-to-date. Anecdote: two years ago a party I was with got stuck in the country side. In the dark. It was pouring. On Christmas. On a, as it turned out, private road. It was only thanks to OpenStreetMap we could actually tell the road we were stuck on, to call for help. According to Google Maps we were in the middle of a large field, it draw an absolute blank there - quite literately; there simply wasn't anything nearby on the map.

Re: [Announce] Poor Maps
 

So, how I can use openstreet maps, in this app?

Re: [Announce] Poor Maps
 

@iHusky: probably all map providers in Poor Maps use OpenStreetMap in one or another form. So, you are using it already.

If your area is not covered, you could ask local government to upload the data into this free database or look into how can you improve it yourself.

Re: [Announce] Poor Maps
 

Let me start off by apologizing for my ignorance, but how do I set up voice navigation with Poor Maps. I read the note on Warehouse which directs users to Mimic (with a Warehouse link). I installed that dutifully. However, now I am wondering, how do I make Poor Maps talk because it has been awkwardly silent. :cool:

On a completely different topic: Am I the only one who thinks the posting on top of this one (#144) is somewhat strange? :confused:

Re: [Announce] Poor Maps
 

1. Go to "Navigation" from the main menu, make sure you select the correct language
2. Once the route is found, tap the navigation block at the top of the screen and select "Voice navigation instructions"
3. Check your device's volume level

There shouldn't be more it than that.

Re: [Announce] Poor Maps
 

Thanks, that confirmed my settings after trying for hours on my "testing" Jolla 1 phone and finally just repeating these steps on my "production" Jolla 1 phone: While immediately working on the "production" phone, the "testing" phone is still silent.

I am using picotts under SFOS 2.1.3.7
pico2wave --wave=foo.wav "$(cat foo.txt)" is working fine on both phones.
BTW, mimic -v foo.txt foo.wav generated a garbled vocalisation and consumed a third of the available processing power while doing so, hence IMO picotts is preferable on Jolla 1 phones.

Any idea how to debug PoorMaps not using picotts on my "testing" phone?

Re: [Announce] Poor Maps
 

@olf, assuming that mimic is installed correctly and didn't get damaged on download, there most probable cause of problems is OOM killer.

In general, Poor Maps has preference among TTS engines which is

mimic [best]
flite [2nd]
picotts
espeak

The first on the list gets picked. So, if you have mimic installed, that would be used.

Now, if OOM gets involved, you could test it by checking what's exit value after running in shell. Mimic has several voices with the default one the most complicated (male voice ap). So, if my suspicion is right, you could switch to a simpler model by choosing female voice (slt) in Poor Maps preferences and see if it will start working.

I doubt that mimic is using any fancy hw-acceleration that could hit some other limits than RAM in J1. Since you have multiple J1, you can always run a small cluster of them and share the tasks around (would require some coding on Poor Maps side). While a joke, its surely doable :) .

I hope that switching the voice will resolve your problems. If not, please try to figure out whether OOM killer was involved.

Re: [Announce] Poor Maps
 

Thanks @ringius for your reply, but unfortunately it misses the point a bit:

Well, pckon / rpm should have complained then, right?

I did not watch or capture memory statistics, but mimic was eating 1/3 of the available processing power of the Jolla 1 phone (i.e. most of one core), when used on the command line (in fingerterm, with no other applications running, but AlienDalvik started).
Hence I concluded that using mimic is not a good choice on Jolla 1 phones, as it is supposed to run concurrently with OSMscout server and a native Map app, plus eventually other apps. If mimic's memory consumption is known to be excessive, that supports this conclusion.

Thanks, I have not found that clearly documented elsewhere yet, but already guessed so.

AFAIR it was 0, i.e. O.K.

But for these technical reasons (and as mimic is English language only), I uninstalled mimic some days ago and decided to use picotts (as pointed out in my last post in this discussion thread).

Still I wonder, why picotts works flawlessly within PoorMaps on one Jolla 1 phone and is silent on the other (while pico2wav is working fine, if invoked at the shell).
Any hints on how to debug this are appreciated, still.

Re: [Announce] Poor Maps
 

Run Poor Maps from the command line ("sailfish-qml harbour-poor-maps"), the voice generation commands and process return values should appear in the console output.