maemo.org - Talk - [SailfishOS] [Announce] Poor Maps

Page 2 of 4

Show 40 post(s) from this thread on one page

maemo.org - Talk (https://talk.maemo.org/index.php)

- Applications (https://talk.maemo.org/forumdisplay.php?f=41)

- - [SFOS] [Announce] Poor Maps (https://talk.maemo.org/showthread.php?t=96781)

eson	2017-01-06 18:03

Re: [Announce] Poor Maps

Quote:

Originally Posted by otsaloma (Post 1521518)

The translations in your first screenshot do work, the problem is probably in whichever way you tested it. For a proper test, you should have Qt Linguist and gettext installed, run "make rpm", which also generates the translations files from po files, and install the resulting RPM. This might not work on the device (gettext too old?), I make the RPMs on my desktop Linux system.

Thanks! I'll try make on my desktop.

Quote:

The time remaining ("1 h 34 m / 1 h 34 m") is used for brevity. "1 h 34 min / 1 h 34 min" would be split across two lines, at least on the Jolla 1 screen. I'm open to better suggestions.

"h" and "m" is absolutely okay, I just wanted to translate to the Swedish equivalent (t and m). Other languages might have other equivalents.

Quote:

TThe navigation instructions come in most cases from the provider APIs. The providers don't support all languages. Mapzen supports cs-CZ (cs), de-DE (de), en-US (en), en-US-x-pirate (pirate), es-ES (es), fr-FR (fr), hi-IN (hi), it-IT (it) and MapQuest Open en_US, en_GB, fr_CA, fr_FR, de_DE, es_ES, es_MX, ru_RU. OSRM is English only and Digitransit translated as part of Poor Maps.

Sadly, just as I suspected. :(

carmenluci

2017-01-06 20:07

Re: [Announce] Poor Maps

Quote:

Originally Posted by eson (Post 1521520)

Sadly, just as I suspected. :(

Hi eson! If you use OSM Scout Server and translate it to Swedish language, you will get full translation in Poor Maps ;)

eson	2017-01-06 20:37

Re: [Announce] Poor Maps

Quote:

Originally Posted by carmenluci (Post 1521526)

Hi eson! If you use OSM Scout Server and translate it to Swedish language, you will get full translation in Poor Maps ;)

Thanks! I will do that. I'm actually halfway through that translation. :)

eson	2017-01-07 10:15

Re: [Announce] Poor Maps

Quote:

Originally Posted by eson (Post 1521527)

Thanks! I will do that. I'm actually halfway through that translation. :)

Hmm... American road classification is really hard to transfer into European conditions. :(
Ideas anyone?

Edit:
Think I found something. This might be helpful also for other translators...
http://wiki.openstreetmap.org/wiki/Map_Features#Highway
.

atlochowski

2017-01-07 17:38

Re: [Announce] Poor Maps

Quote:

Originally Posted by eson (Post 1521539)

Think I found something. This might be helpful also for other translators...
http://wiki.openstreetmap.org/wiki/Map_Features#Highway
.

Thanks. I got stuck on this kinds of roads some time ago. It will help to finish translation.

otsaloma

2017-01-12 21:31

Re: [Announce] Poor Maps

Quote:

Originally Posted by eson (Post 1521520)

"h" and "m" is absolutely okay, I just wanted to translate to the Swedish equivalent (t and m). Other languages might have other equivalents.

Time strings are now translatable. The shortening remains not translatable, it's always to the first letter of the unit label. New strings at Transifex.

https://github.com/otsaloma/poor-map...0ffee17770de2b

Polish and Russian translations are incomplete. I could wait a while if someone wants to finish those and then make a release.

otsaloma

2017-01-21 17:19

Re: [Announce] Poor Maps

0.28.1 released with German (Germany), Spanish (Spain), French, Dutch and Swedish translations. Thanks to all translators. I left out the incomplete Polish, Russian and Italian translations.

Transifex doesn't seem to show comments prominently, but translators please note that "translator-credits" should be translated to your own name, it will show up in the about dialog. If you want to remain anonymous, then of course you can use "Anonymous" or "Translator" or something.

claustn

2017-01-22 10:01

Re: [Announce] Poor Maps

I finished the italian translation, it should be overall ok.

monkeyisland

2017-02-09 20:45

Re: [Announce] Poor Maps

Hello there since last jolla update 2.1.0.9 poor maps cant access maps from OSM Scout Server now.

http://s26.postimg.org/io5fopryx/Scr...170209_002.png

And on Navigation settings a change does nothing.... in this example
i tried to click on OSM

http://s26.postimg.org/uihs03fmh/Scr...170209_003.png

http://s26.postimg.org/9zmvv11p5/Scr...170209_005.png

http://s26.postimg.org/vakfzajtl/Scr...170209_006.png

:confused:

otsaloma

2017-02-09 20:49

Re: [Announce] Poor Maps

I'll probably upgrade on the weekend and take a look then. I'd expect some amount of breakage due to the switch to Qt 5.6.

otsaloma

2017-02-12 01:53

Re: [Announce] Poor Maps

Bad news: the new version of PyOtherSide included in Iijoki seems significantly broken, at least with respect to how I'm using it. It's somewhat difficult or at least seriously inconvenient to work around, so it might take a while to fix this.

https://together.jolla.com/question/156736/

Could someone still using the previous version of SFOS download and share the RPM of "pyotherside-qml-plugin-python3-qt5"? Downgrading to that could be a temporary work-around.

There might be other problems with Iijoki as well, but I'd need this fixed before I can proceed.

Jordi

2017-02-12 07:26

Re: [Announce] Poor Maps

1 Attachment(s)

Here you go.

otsaloma

2017-02-12 14:04

Re: [Announce] Poor Maps

Thanks, but that 1.4.4 version didn't seem to work either. Probably something else in the Iijoki update has triggered the bug.

But anyway, on second thought, it seems fairly easy to work around this after all, in the common case of scalar arguments and assuming no weird values. For anyone else interested, something like the below seems to work.

Code:

Python {

    id: py

    ...

    function call_sync(func, args) {

        // XXX: Work around a call_sync bug by using evaluate.

        // https://github.com/thp/pyotherside/issues/49

        args = args.map(py.stringify).join(", ");

        return py.evaluate("%1(%2)".arg(func).arg(args));

    }

    function stringify(obj) {

        // Return Python string representation of obj.

        if (Array.isArray(obj)) {

            return "[%1]".arg(obj.map(py.stringify).join(", "));

        } else if (obj === null || obj === undefined) {

            return "None";

        } else if (typeof obj === "string") {

            return "'%1'".arg(obj.replace(/'/, "\\'"));

        } else if (typeof obj === "number") {

            return obj.toString();

        } else if (typeof obj === "boolean") {

            return obj ? "True" : "False";

        } else if (typeof obj === "object") {

            // Assume all remaining objects are dictionaries.

            return "{%1}".arg(Object.keys(obj).map(function(x) {

                return [py.stringify(x), py.stringify(obj[x])].join(": ");

            }).join(", "));

        } else {

            throw "Unrecognized argument type: %1: %2"

                .arg(obj).arg(typeof obj);

        }

    }

}

otsaloma

2017-02-15 22:11

Re: [Announce] Poor Maps

0.28.3 available at OpenRepos. I didn't notice any bugs related to Qt 5.6, only that above PyOtherSide bug. In addition to the fix, Italian and Polish translations added.

Feathers McGraw

2017-02-15 23:21

Re: [Announce] Poor Maps

Quote:

Originally Posted by otsaloma (Post 1523533)

0.28.3 available at OpenRepos. I didn't notice any bugs related to Qt 5.6, only that above PyOtherSide bug. In addition to the fix, Italian and Polish translations added.

Have you changed the vendor name in the rpm? I get this when trying to update with zypper (packagekit won't update it):

Code:

There is an update candidate for 'harbour-poor-maps', but it is from different vend

or. Use 'zypper install harbour-poor-maps-0.28.3-1.noarch' to install this candidat

e.

otsaloma

2017-02-15 23:35

Re: [Announce] Poor Maps

Quote:

Originally Posted by Feathers McGraw (Post 1523536)

Have you changed the vendor name in the rpm? I get this when trying to update with zypper (packagekit won't update it):

I don't even know what "vendor" means. There's no vendor in the spec file or in "rpm -qpi" output. I haven't changed anything in the packaging. Sounds like some zypper oddity.

Feathers McGraw

2017-02-16 09:20

Re: [Announce] Poor Maps

Quote:

Originally Posted by otsaloma (Post 1523539)

I don't even know what "vendor" means. There's no vendor in the spec file or in "rpm -qpi" output. I haven't changed anything in the packaging. Sounds like some zypper oddity.

Not a zypper oddity, zypper is just giving more information. Trying to update with e.g. warehouse was giving "no update candidate".

There might not be a vendor now, but there used to be! It was meego for the older packages (at least, that's what zypper said when I forced an upgrade: changing vendor "meego -> ").

Perhaps something in the SDK or build environment used to add meego as vendor if none was specified, and you're building in a new SDK which doesn't do that any more?

It's not a problem because you can always uninstall and reinstall, or do what I did and force the update using zypper, but people are unlikely to understand why it won't update.

P@t	2017-02-16 09:45

Re: [Announce] Poor Maps

Quote:

Originally Posted by otsaloma (Post 1523539)

I don't even know what "vendor" means. There's no vendor in the spec file or in "rpm -qpi" output. I haven't changed anything in the packaging. Sounds like some zypper oddity.

in case that helps, there is a section in https://harbour.jolla.com/faq about the 'vendor' issue...

claustn

2017-02-16 11:50

Re: [Announce] Poor Maps

Does the new version work on 2.0.4 ? I would like to check my translation but I can't update it through Warehouse

Feathers McGraw

2017-02-16 12:13

Re: [Announce] Poor Maps

Quote:

Originally Posted by claustn (Post 1523563)

Does the new version work on 2.0.4 ? I would like to check my translation but I can't update it through Warehouse

Same problem as I had? You can update using this command in zypper:

Code:

devel-su zypper install harbour-poor-maps-0.28.3-1.noarch

If you don't have zypper installed, install it first with:

Code:

pkcon install zypper

monkeyisland

2017-02-16 15:39

Re: [Announce] Poor Maps

I installed lasted version from openrepos.
I dont know if the issue comes from poor maps or osm scout server.
Now i can choose maps from osm. But when i try navigate with osm scout in poor maps osm scoit server will crash.
Offline search is also not possible.

rinigus

2017-02-16 16:21

Re: [Announce] Poor Maps

Quote:

Originally Posted by monkeyisland (Post 1523581)

In general, if OSM Scout Server crashes - its a problem with the server, not Poor Maps. I'll reply in more details in OSM Scout Server thread to avoid info duplication

otsaloma

2017-02-16 20:42

Re: [Announce] Poor Maps

Quote:

Originally Posted by Feathers McGraw (Post 1523553)

There might not be a vendor now, but there used to be! It was meego for the older packages (at least, that's what zypper said when I forced an upgrade: changing vendor "meego -> ").

Perhaps something in the SDK or build environment used to add meego as vendor if none was specified, and you're building in a new SDK which doesn't do that any more?

Maybe you installed previously from Mer OBS? That seems to add the meego vendor.

https://build.merproject.org/package...mmon/poor-maps

otsaloma

2017-02-16 20:44

Re: [Announce] Poor Maps

Quote:

Originally Posted by claustn (Post 1523563)

Does the new version work on 2.0.4 ? I would like to check my translation but I can't update it through Warehouse

No, you need to upgrade to SFOS 2.0.5 or later.

Feathers McGraw

2017-02-16 21:02

Re: [Announce] Poor Maps

Quote:

Originally Posted by otsaloma (Post 1523605)

Maybe you installed previously from Mer OBS? That seems to add the meego vendor.

Not knowingly, I originally installed from openrepos, but I am using FP2 which has a few more repos so I guess it could have been updated when I updated to 2.1.

The rpm version I upgraded from had ".jolla" at the end like the one on OBS, too.

If that is what happened, it's weird that it updated to the OBS version without complaining about the change of vendor, maybe it doesn't see a conflict if it used to be set to an empty string and now has a value. Or during release mode it might override the default behaviour.

bomo	2017-06-16 19:08

Re: [Announce] Poor Maps

Sorry for being potentially annoying, but is there any (rough) ETA for https://github.com/otsaloma/poor-maps/issues/26 and/or https://github.com/otsaloma/poor-maps/issues/30?

otsaloma

2017-06-17 11:55

Re: [Announce] Poor Maps

Quote:

Originally Posted by bomo (Post 1529246)

Sorry for being potentially annoying, but is there any (rough) ETA for https://github.com/otsaloma/poor-maps/issues/26 and/or https://github.com/otsaloma/poor-maps/issues/30?

No ETA, not even rough. I can tell you that I'm prioritizing the tile server matter ahead of those and will start on that maybe in two weeks or so. Tile servers are currently not an issue, but have potential to become a big problem.

https://github.com/otsaloma/poor-maps/issues/33

rinigus

2017-06-17 16:53

Re: [Announce] Poor Maps

Quote:

Originally Posted by otsaloma (Post 1529274)

This can be, indeed, a huge issue for all using online servers. How do you envision rolling out your own service? Do you plan to enrol CDN and servers or is there any other way? Compared to just distributing of the files, map tile generation requires CPU and that maybe significant expense.

Looks like Klokantech also provides maps access via selling it to individual users, if I read the end of the page http://www.tileserver.com/ correctly. Maybe there are some other services were you could offload handling users directly.

otsaloma

2017-06-17 18:53

Re: [Announce] Poor Maps

Setting up a tile server would mean getting some kind of cloud server resources. In addition to CPU, probably large fast SSD-drives are needed. I don't know the specifics yet, but considering the commercial pricing (e.g. Mapbox's $500/month), and considering that it doesn't look too difficult these days, I'm going to give it a try. It's well possible that it turns out to be too expensive or too slow at affordable prices, but it's worth a look.

rinigus

2017-06-18 07:38

Re: [Announce] Poor Maps

Quote:

Originally Posted by otsaloma (Post 1529285)

[I am writing it down here just as a note that would hopefully help you when you start looking into it.]

500/month is rather steep. When looking for CDN, I found comparison on https://www.cloudorado.com/ rather helpful. At least, it allowed me to see the cheaper CDN providers and look later on their offers. In addition to CPU and storage (probably around 100+GB for planet or a bit more), look for transfer out which could easily become a significant expense. Here, the bandwidth provided by data.modrana.org has helped and made it possible to use without collecting money.

Do you have any idea of the budget?

Looking on all complications, maybe https://openmaptiles.com/hosting/ is the option worth considering. But again, it would would require some mechanism allowing users to pay for it.

bomo	2017-07-22 18:25

Re: [Announce] Poor Maps

Are there also plans to add a POI search? like listing all POIs of a particular kind in around an address?

Another question: Is it planned to add a feature that allows to click on POI icons shown on the map and view details of it?

otsaloma

2017-07-22 19:51

Re: [Announce] Poor Maps

Quote:

Originally Posted by bomo (Post 1531111)

"POI search" already exists. It's called "Nearby venues". You can access it from the main menu or a search result bubble on the map.

The maps are raster maps, so, no you can't interact with any POI icons there. What you can do is a nearby search and from the search result list, choose "Show all" from the pulldown menu. After that you'll have a bunch of blue circles on the map that are clicable. The Foursquare provider gives some details of places, others only the name.

otsaloma

2017-08-26 15:53

Re: [Announce] Poor Maps

Version 0.32 available at OpenRepos.

Add support for rerouting (#26, rinigus)
Redesign navigation status and narrative pages
Add OSM Scout car day and night maps (rinigus)
Allow searching by latitude/longitude coordinates, e.g. "60.169 24.941" or "60.169,24.941" (period as decimal, any non-alphanumeric separator between latitude and longitude)
Fix auto-centering in landscape in navigation mode to not have the position icon overlap with the menu button (#46)
Do relative requests to download tiles, fixes HSL map
Fix string escaping in Python calls from QML (#50, rinigus)
Add Russian translation
Update translations

Thanks to rinigus we now have support for automatic rerouting. This works best with the Valhalla routers (Mapzen Turn-by-Turn and OSM Scout) as they will account for the heading and avoid suggesting immediate U-turns.

Translations that were broken in the earlier early access Sailfish OS 2.1.1.x work again with 2.1.1.26.

bomo	2017-08-28 08:17

Re: [Announce] Poor Maps

Thanks a lot! I today tried the automatic rerouting functionality and it's working fine. May I ask whether it is possible to add an option to adjust the details of when it starts the rerouting ? Would it also be possible to add a 'lock on streets' functionality if I select car navigation ? Or is that not possible due to the maps/tiles ?

rinigus

2017-08-28 10:39

Re: [Announce] Poor Maps

Quote:

Originally Posted by bomo (Post 1533411)

@bomo, maybe you could be a bit more specific:

* Re rerouting details: which details do you want to adjust? In general, its all expected to be working out of the box, without any adjustments needed. To know the internal parameters - which router did you use in Poor Maps (Mapzen, OSM Scout, ...)?

* Re lock on streets: I presume you mean something like shown as a map matching service at https://mapzen.com/blog/map-matching/? Its possible to implement it, would need some time though. The router behind Mapzen and OSM Scout (Valhalla) does support map matching and we could implement it for online and offline navigation. "Just" have to expose functionality, connect it to Poor Maps, and allow users to enable/disable it.

I wonder whether you have terrible GPS signal and rerouting is triggered too frequently?

bomo	2017-08-28 17:14

Re: [Announce] Poor Maps

Quote:

Originally Posted by rinigus (Post 1533426)

Thanks for your reply.

@1: I was mainly thinking of the deviation parameter (start rerouting after x meters deviation from route). I was using poor maps with Mapzen navigation. So basically, I also had the impression the rerouting was triggered too late.

@2: Yes, that's exactly what I mean. Sounds great!

Another question: Currently, in the navigation mode, the map is rotated according to my orientation (which is perfect), but the street names on the map etc. are not rotated. Is there also a possibility to change this?

otsaloma

2017-08-28 18:45

Re: [Announce] Poor Maps

The rerouting trigger is currently 200 meters (plus positioning accuracy), polled every three seconds. 200 meters is about a block and a half in a common city center, or way less than motorway exit spacing. So, how many blocks late do you see rerouting happening? What's your positioning accuracy (see bottom of the screen, below the speed)? You're not expecting to catch the very next turn, are you? Do you have experience with Google/Apple/HERE maps, what's the trigger there? I don't want to add an option, there should be no need for such, but we'll of course change the default if it's bad.

Quote:

Originally Posted by bomo (Post 1533461)

Another question: Currently, in the navigation mode, the map is rotated according to my orientation (which is perfect), but the street names on the map etc. are not rotated. Is there also a possibility to change this?

No, not with raster maps, and it's too early to talk about vector maps.

bomo	2017-08-28 20:39

Re: [Announce] Poor Maps

Quote:

Originally Posted by otsaloma (Post 1533464)

Thanks for the clarification. I need to check again the accuracy etc. I also try to make a screenshot next time.

I dont know about google etc, but osmand seems to use 100m limit.

rinigus

2017-08-29 08:02

Re: [Announce] Poor Maps

Quote:

Originally Posted by bomo (Post 1533475)

Thanks for the clarification. I need to check again the accuracy etc. I also try to make a screenshot next time.

I dont know about google etc, but osmand seems to use 100m limit.

I guess we should try 100 meters and see if its OK.

Note that rerouting distance is a compromise. When you use offline router (like OSM Scout), you are using your own device and the main price to pay is the battery. When you use online router, all API calls by all users are summed up by the provider and if we exceed the free tier we would have to find the way to pay for it. Hence having too small rerouting distance with online routers could lead to problems. @otsaloma has a better overview and he'll let us know if we are going to hit it, I presume.

As for map matching, I added it into OSM Scout Server issues and will look into it when I get there.

Rotating labels are possible, in theory, using tiles as well. For Mapnik, it would mean composing other projection that would flip the world. However, it would require significant time to do that properly. As mentioned by @otsaloma, an alternative is to look into vector maps. In long term, vector maps (OpenGL-based rendering) are better solution and I probably would prefer to look into that than spend too much time on hacking Mapnik projections.

otsaloma

2017-08-29 17:38

Re: [Announce] Poor Maps

100 meters is possible, yes. But, I'd still like some verification that it is indeed the 200 meters part at fault, so that we'd be fixing the right thing.

Quote:

Originally Posted by rinigus (Post 1533497)

Note that rerouting distance is a compromise. When you use offline router (like OSM Scout), you are using your own device and the main price to pay is the battery. When you use online router, all API calls by all users are summed up by the provider and if we exceed the free tier we would have to find the way to pay for it. Hence having too small rerouting distance with online routers could lead to problems. @otsaloma has a better overview and he'll let us know if we are going to hit it, I presume.

Mapzen's pricing is very reasonable and once past the free plan the prices are per request, so there's no big step like Mapbox's $500/month. So, no need to worry about my cash flow. But, it is of course possible for some users to have high data traffic costs, maybe abroad especially.

Besides cost, two other reasons come to mind for not making rerouting too sensitive: (1) I don't know if the positioning accuracy is always guaranteed to be correct and in sync -- e.g. what if when entering a tunnel your positioning throws quickly while the accuracy lags? And, (2) there are normal small deviations, such as lane choice on big motorways or intersections, or detours around road works, which can be maybe up to 50 meters.

Quote:

Originally Posted by rinigus (Post 1533497)

Rotating labels are possible, in theory, using tiles as well. For Mapnik, it would mean composing other projection that would flip the world. However, it would require significant time to do that properly. As mentioned by @otsaloma, an alternative is to look into vector maps. In long term, vector maps (OpenGL-based rendering) are better solution and I probably would prefer to look into that than spend too much time on hacking Mapnik projections.

With a custom tileserver and custom requests that include the heading, yes, it could be done. But, then you'd have a cache invalidation problem and would need to regenerate a lot of tiles as the heading changes. Doable, yes, but indeed more trouble than it's worth.

All times are GMT. The time now is 13:34.

Page 2 of 4

< Prev

Next >

Show 40 post(s) from this thread on one page

vBulletin® Version 3.8.8