Contribute to Open Source. Search issue labels to find the right project for you!

Ensure sub-processes and resources work in production


From #2:

  • Other sub-processes/resrouces to still be tested:
    • ActiveMail - SMTP
    • ActiveJob - Redis
    • ActionCable - Redis, CORS
    • Cache - Redis

Some things we might need: - Dockerfile.sidekiq - Dockerfile.cable so we can do (see this doc):

heroku container:push --recursive
=== Building web
=== Building sidekiq
=== Building cable

Also, consider renaming website to web in docker-compose.yml to align with the web Dyno process.

Updated 20/08/2017 04:11

Headers in lowercase, including bearer

Bug or feature request? Bug
Which Swagger/OpenAPI version? 2.0
Which Swagger-UI version? 3.1.5
How did you install Swagger-UI? download of dist folder
Which broswer & version? Chrome Version 60.0.3112.90 (Official Build) (64-bit)
Which operating system? Mac OS 10.12.6

Demonstration API definition

API with OAuth2 support

Configuration (browser query string, constructor, config.yaml)


Expected Behavior

I authorized swagger-ui using the default settings, against my OAuth 2 server, with the “implicit” grant.

When using “try it out”, it sends the following headers: -H "accept: application/json" -H "authorization: bearer ..."

At least the “bearer” part of this should be uppercase:

Also, I think it’s more standard to send title-case headers, although they’re technically case-insensitive.

Current Behavior

“bearer” is lowercase.

Possible Solution


Updated 18/08/2017 18:14 2 Comments

Content Editing UI


If there’s time, we could figure out a way to create a little CMS for the site.

Since we’re generating the site from JSON, we could use to generate it from the JSON schema. This would be super simple, the only challenges are:

  • [ ] how to make it private, yet available to editors, i.e. where to host it
  • [ ] how to trigger the page update (AWS Lambda?)
Updated 15/08/2017 14:08

Textured Backgrounds


This is a separate ticket because we can do some interesting things to improve loading perf.

According to Nick:

  • the background shapes are vectors, so they can be loaded as tiny SVGs
  • the textures could be very low bitrate PNGs (they could probably just use 2 colours, making them only up to a few KB)

Which means we could first load the SVGs, put coloured backgrounds underneath to get the plain colours, and then load the texture PNGs progressively.

Updated 15/08/2017 13:48

Ocultar los mensajes de object


Sobre todo en las primeras clases, que al desplegar el menu contextual de los mensajes disponibles en un wko muestre tambien los mensajes de object, entorpece mas que ayudar. Despierta preguntas que desvian la atención y no hay como responder sin espoilear.

Con el mismo criterio de no mostrar object en el diagrama estatico, ni necesitar hacer inherits para heredar de object para no anticipar conceptos, seria mejor ocultarlos. A lo sumo, permitir configurar para que un usuario avanzado lo habilite.

Aparte, excepto el == y el \=, el resto dificilmente se usa.

Updated 10/08/2017 23:07

Mapillary visually covers Names and POI icons


OsmAnd+ 2.7.4, released 2017-07-01 Samsung GT-I8190 running Andorid 4.1.2

There seems to be an unfavourable illustration issue with Mapillary. When Mapillary is activated, its green lines visually cover names of citys, regions, streets and even POIs.

Suggestion: Place the Mappilary layer in the back so you can actually see the mentioned names and POI icons at any time in front. If one would like to access a mapillary image in a certain spot/street he/she may has to magnify the map view to a certain extend at which e.g. the POI doesn’t visually block the Mapillary dot anymore


Updated 05/08/2017 10:14 2 Comments

Make react imports easier


Right now, the directory structure of the frontend app has top level directories api, assets, and components. Getting access to files in other top-level directories is a fragile process of inserting ../ until your import works, which is not great. There’s a way to make eg. api/user_api' work as an import path, but I need to look into that.

Updated 04/08/2017 04:18

German TTS skipping latvian letters


Hello! I am using the newest F-Droid build of Osmand (also bought it on the Playstore just to support this great app!). When reading, the German TTS simply skips latvian special letters: ē, ī, ū, ā. Result for example: “Rīga” will be pronounced “Rga”; making many towns/street names unrecognizable. I have Pico TTS activated. I tried another app, utilizing pico TTS, which is able to hand over these letters to Pico TTS, which pronounces them (in a german way obviously) mostly correctly. So i guess the bug is that the german Osmand TTS plugin just doesn’t hand them over to pico, as they are not German letters.

Thank you!

EDIT: Oh. I should not forget the capital letters! Here they are: Ē, Ī, Ū, Ā

Updated 15/08/2017 17:14 2 Comments

POI prioritization on map


I’m using OsmAnd+ 2.6.5 (2017-05-31) with it’s OSM editing plugin.

Sometimes I feel that the map shows too many POIs and sometimes it’s the opposite: I can’t find what I’m looking for because I set the POI filter to a wrong type.

Ideally OsmAnd would have a priority value between 0 and 1 for all POIs, depending on various factors like navigation mode (walking, driving or riding a bike), it’s distance from the user, is it open now or not, or special needs like wheelchair access or acceptance of credit cards or cryptocurrencies.

Currently OsmAnd let’s us filter POIs by type, or use the Nearest POIs option. This is too binary I think, and a customizable fuzzy logic would work much better.

For example: When you are riding a bike, you would potentially need to know where the free water taps, petrol stations with compressor, bike repair shops, open restaurants, cafes, shops, tourist attractions are, but you don’t really need to know about others. While driving you might be interested in fast food drive-throughs, all petrol stations, restaurants, free parking spaces etc. Users with special needs like wheelchair users or crypocurrency enthusiasts would choose those places to show it they are available in close proximity.

So the idea is that there would be preset priority tables for the navigation modes, where we could define a priority numbers (-10.0 … +10.0, -10 = very important and prefer not to have it, 0 = not important at all, +5 = normal priority, prefer to have it, +10 = very important, prefer to have it) for POI tags. Every POI tag would have the value of 0 (ignore tag score) by default, but the car/bike/walking presets would customize these values. The POI tags also would get a score between 0 and 100. The scores would be calculated differently for different tags. There would be some obvious one like wheelchair access (yes = 100, no = 0, limited = 50) or internet (yes = 100, no = 0). The priority of the POI would be calculated by adding all it’s tags' priority numbers and their scores, divided by the number of tags in the equation. This way a POI would always get a priority score between -1000 and +1000.

For example: Ankert in Budapest, Hungary is a dog-friendly pub with wifi internet and wheelchair access. It closes at 2:00. Kadarka Kóstoló Bár is a pub close by, but it’s not wheelchair accessible and it closes at midnight.

By default these two places would have very similar priorities, if we are driving close to them. The situation would change, if it was 1:00 in the morning: Kadarka would get a penalty, it’s priority score would be smaller, because it’s closed at the moment (opening_hours tag’s priority normal [+5], it’s score 0) compared to Ankert (opening_hours tag’s priority normal [+5], it’s score 100). The preset priority table would be combined with the user’s priority table, like wheelchair access (-10 - I don’t want to have it, -5 - I prefer not to have it, 0 - I don’t mind if the place has it, +5 - I prefer if it has it, +10 - I really need wheelchair access) or credit card acceptance, etc. If the user defined that he/she must have wheelchair access, Kadarka would be not shown on the map at all. If the user defined that he prefers dog-friendly places, Ankert would get a higher priority score, but Kadarka’s priority would not change, since we don’t have any information about its dog-friendliness.

So mathematically every POI would get a priority score by summing all its tags' priorities and their scores, and this POI priority score would change the size of the POI and/or we could hide POIs below a certain priority value depending on the zoom level on the map. If you want to have more results on the map you could adjust the threshold above which priority score should a POI be displayed.

It probably sounds complicated, but I hope you can see some value in it, I think it has the potential to improve the search and POI display significantly.

Updated 03/08/2017 10:28 1 Comments

Bundle test script


Right now the packaging is a bit broken: The lcdtests package is packaged too, but without the script.

We should probably create a separate package (called RPLCD_test or something like that) that will be bundled with the main library and make sure that the test script is automatically installed in the PATH for the user, so that the tests can be run after a pip install from PyPI too.

Updated 30/07/2017 22:41

Option "Avoid stairs / dirt roads" for pedestrians too


In the BIKE routing options one may “avoid stairs” (“Treppen vermeiden” in German) and “avoid dirt roads” (“Unbefestigte Wege vermeiden” in German).

In PEDESTRIAN routing settings these options are missing, although they are similarly important for many people. Consider e.g.:

  • parents moving a pram (“Kinderwagen / Kinderkarre”)
  • people using crutches / walking sticks / prostheses (“Krücken / Gehhilfen / Prothesen”)
  • persons in wheelchairs (“Rollstühlen”)

I don’t want to open the “can of worms” to fully support people with whatever sort of handicap / feature.

But the algorithm / code to avoid these 2 barriers is already existent in bike mode, isn’t it? Just enable the option settings for pedestrians too.


Updated 01/08/2017 12:12

Underlay map: Polygons are shown even though "Show polygons" is off


When the map is zoomed to a certain zoom level, OsmAnd starts to show polygons. This is for the “Underlay map” feature, and the “Show polygons” switch is off. Base map transparency slider is to the right. (Side note: Actually, the slider is base map opacity, not transparency, if we assume that the slider increases towards the right). The map example is from lat, lon: 59.94889, 10.76950,

screenshot_2017-07-30-1 Underlay map: Statkart topo (NO). Show polygons: off

screenshot_2017-07-30-2 Polygons are not shown, everything is ok.

screenshot_2017-07-30-3 Zoomed in one more level. Polygons are shown, despite the setting being “off”. Continue zooming, and polygons are still shown.

Versions: OsmAnd+ 2.7.3, released 2017-07-29 Offline map of Norway/Oslo, standard map, 2017-07-01 Device: Samsung Galaxy Note 8.0 GT-N5120, Android version 4.4.2

Updated 05/08/2017 11:36 2 Comments

Spanische Übersetzung muss erneuert werden


Es sind ein paar Strings hinzugekommen oder wurden geändert. Daher bitte bei Gelegenheit die spanische Language Datei language/es_ES.po und im dev-1-7- Tree updaten.

(Am einfachsten mit dem Plugin Loco Translate auf einen geeigneten Testserver).

In dem Zusammenhang wäre ein Blick in die en_GB und die en_US auch nice. Derzeit ist en_US eine Kopie der en_GB. en_US haben wir nur noch zur Abwärtscompatibilität. Ich würde es auch löschen, aber dann haben einige Websites, die auf en_US gesetzt haben statt auf en_UK Probleme….

Updated 28/07/2017 21:33 2 Comments

Text-to-speech should pronounce road numbers depending on their length


This is referring to the ‘en-GB’ voice, although the standard ‘en’ voice produces the same results.

The usual practice within the UK is to pronounce road numbers with one or two digits as their natural number, but to spell out numbers longer than that as digits. For example, someone would say the name of the A1 or the A57 as “A one” or “A fifty seven” respectively, but would say A307 as “A three oh seven” or B3152 as “B three one five two”. Currently the TTS says the latter two as “A three hundred and seven” and “B three thousand one hundred and fifty two” respectively.

I’m not sure what is normally done in other countries or with other languages, but saying road numbers in this way (if it is technically possible) would sound more familiar and aid with reading road signs.

Updated 05/08/2017 10:07 3 Comments

Support for WMS/WMTS


Trying to add the online vector map from Lantmäteriet, the Swedish National Land Survey, but can’t make it work. Probably because it is WMTS and Osmand just support TMS, or am I wrong?

Don’t know anything about technical difficulties but would it be possible to add a support for WMS and WMTS?

Updated 01/08/2017 12:14

create a map with all features as legend


hi there! just an idea that crossed my mind when trying to find how to decrypt the colors of hiking path… is it possible to create a map that contains all or most of the features displayed on a map? if possible a map that can be used as overlay or underlay of the map…. so that we can switch to it fast…. when walking a map without legend is almost useless…. could be great also to be able to select what should be rendered on the map, it will speed up rendering while keeping all useful information for a particular use…. wandering path are useless in bicycle, but some times you want them to be displayed…. a screen with |——drawing———-|——–description——|—–enable checkbox—-| would be the top feature!!!! thanks and regards! JLM

Updated 25/07/2017 07:53

Define expression to (road name) pronunciation


I’m using OsmAnd+ 2.6.5 (2017-05-31) with it’s OSM editing plugin.

I think it is more like a general, Open Street Map feature request. There are many primary roads, highways named with numbers. When the TTS engine generates their names and puts them into the instruction they sound unnatural.

For example: we have a primary road called “5” in Hungary. When OsmAnd generates an instruction to turn right onto that road, it says something like this: In two hundred meters turn right here: five.

The ideal sentence would be more like: In 200 meters turn right onto the primary road (number) five.

So, ideally we could define a local, language specific expression for road types to generate “pronunciation names” if the name of the road is only a number, or something like M5 (as for motorway number 5).

Updated 25/07/2017 08:16 2 Comments

Combine two TTS engines to pronounce foreign names corretly


I’m using OsmAnd+ 2.6.5 (2017-05-31) with it’s OSM editing plugin.

Text-To-Speech engines are getting better and better, and OsmAnd uses them wonderfully. Using them abroad gets funny though, when the TTS tries to pronounce the street names in a non-local language.

Theoretically it would be possible to use the user-preferred TTS engine to generate instructions, but use a different, local TTS to generate the name of the streets. This way the tone of the street name would be different from the one of the instructions, but it would allow the user to learn the local names with the correct pronunciation. This would be useful in many cases, for example when you want to ask for a local for directions.

So my suggestion would be to use two TTS engines in harmony: use one native to the user for the instruction generation, and another for the street names. (Obviously they would be the same if the user is in his/her homeland, or if he/she chooses to use the local language for instructions.)

Updated 25/07/2017 09:06 2 Comments

Add route percentage display


I’m using OsmAnd+ 2.6.5 (2017-05-31) with it’s OSM editing plugin.

We have a small time-displaying label on the screen during navigation what displays the remaining time of the trip. If you tap on it it changes to the expected time of arrival.

It would be great to have a third mode, when it displays the completed percentage of the route. It would make the trip more fun especially while walking/trekking or riding a bike.

Updated 01/08/2017 12:27 1 Comments

Collect ratings of places


I’m using OsmAnd+ 2.6.5 (2017-05-31) with it’s OSM editing plugin.

I wonder if it was possible (technically and legally) to collect and display ratings of POIs if the information is available somewhere else.

According to key:facebook and contact:facebook it is possible to refer outside sources of POIs. I couldn’t find a general discussion about it though, so I’m not sure how to indicate other sites like Yelp, TripAdvisor, Google reviews or others. Many times these references are not on the POIs yet, but combined with #3442 users could add those missing links.

Legally I’m unsure if OsmAnd could display the ratings. Did anyone look into this?

Updated 26/07/2017 08:53

Show 'Near city' information for all search results


Hi, In the current version Osmand shows the information that a search result is located ‘x.y km from [the nearest] city’ for suburbs and hamlets only when searching for their names only. 1. This information is not shown when restricting the search to the POI category and explicitly searching for ‘Suburb xy’ or ‘Hamlet xy’ respectively. 2. This information sometimes might be useful for at least towns (in the surrounding of a well known city) too.

I guess that this kind of look-up mechanism determining the distance to the nearest city consumes computational power and therefore this information is not shown for each location and search result. If that is not the reason I would love to see this ‘near city’ information together with the ‘is in’ information proposed in #4174 for each location and search result. And then maybe the interests of #3327 coult be addressed too.


Updated 01/08/2017 12:36

Add 'is in' look-up mechanism to determine at least state/region/community the location is in


Hi, Please add a kind of ‘is in’ look-up mechanism to determine and display information about in which state, region or community and country a selected location (or at least a city, town, suburb, hamlet etc. POIs) is in. It would be great if Osmand can show e.g. that the selected location ‘Innsbruck’ is not just a city but also a ‘City in Tyrol, Austria’. Or that ‘Villalangua’ is a ‘Village in Aragon, Spain’.

In the latter example ‘Aragon’ and ‘Spain’ are also the names of the maps available for download and therefore indicating which maps cover the autonomous community or the whole country respectively around the selected location. (Maybe an indication that the loop-up process may not be too sophisticated since the information about region and country is already somehow used.)

This kind of information would also help to identify and distinguish several similar/identical spelled search results in different countries, like the information ‘x.y km from city’ already shown for hamlets and suburbs does. For that cases, the existing description should be extended by the proposed ‘is in’ information, e.g. ‘45.1 km from Huesca in Aragon, Spain’.

I think as a first step it would be a good idea to implement the proposed ‘is in’ look-up at least for locations/POIs (like cities, towns, suburbs, hamlets etc.) where currently no address look-up is performed. Later on this look-up information could be handsome for any selected or found location. And maybe this would partially cover the interests of #3327.


Updated 02/08/2017 10:04

Add new interesting metrics



What do you think would be an interesting thing to calculate given the data available in our data model? Some examples

  • What languages do an Organisation’s members use?
  • When do an Organisation’s members code? night/day, weekends?
  • What projects are most starred by organisation members?

Think of one, figure out how to calculate it, and then incorporate a visualisation into the web app.

Updated 21/07/2017 14:45

Add intermediate folder in zip file


Split from #84.

The zip file should always have the intermediate folder so that unzipping is grouped.

On MacOS, if the zip file only contains one folder (ie you choose just Home or just Anouncements etc) then the zip file is unzipped down to just that folder. So you get just Home or just Announcements.

If the zip file contains two or more then it will create the intermediary folder.

Can this always be forced?

Updated 22/07/2017 02:17

Auto update distance for nearing POIs in the POI/favourite list



I have following scenario while driving. 450-km drive, 400 of which is a primary highway. So during this 400-km I’m not interested in turns and map. My interests are nearing favorite petrol stations, speed cameras, places to rest, toll boothes.

Currently I open screen with list of all POIs and favourites and observe approaching points. See attached screenshot. The issue is distance to points is not updated automatically, so I have to close the screen and open it in a few minutes.

Is it possible to update distance automatically?

Second issue: points that already passed are not removed from the list so to find nearing points I have to clean up the list by removing outdated from the list top. img_20170719_022907

BTW, removing the point from the list seems to refresh the points distance in the list.

I’m using 2.6.5 Android version.

Updated 19/07/2017 14:31

expand zoom buttons into (virtual) slider [feature request]



while driving I can use double-finger-touch to zoom, but it is not easy to go back. Sometimes I overzoom and need to precisely find the opposite button. Pinching removes the position of te car, which I would like to avoid.

What about using the right side of the display as an invisible slider, so sliding up zooms in (+), sliding down zooms out (-) as two-finger-tab. Im mean the area including (-) and (+) zoom buttons up to - but not including - to whatsever displays at the top right panel (speed, satellites, …)

So sliding expands the zoom buttons in a perfect understandable way and makes it easy without looking too much at the display for executing the action, as the side of the display is easy found with the hand/finger.

Thanks for looking into this Harvey

Updated 19/07/2017 14:28

Visually differentiate POI type and POI name


When a POI is selected on the map an ‘info card’ pops up with information on the POI. The title on the info card reads “POI Type POI Name”. For a example, Rainbow Falls would be “Waterfall Rainbow Falls”.

Proposed change: Visually distinguish the name and the type. This could be as simple as adding a colon, “Waterfall: Rainbow Falls”. Or the name could be on a different line or in a different font.

Updated 01/08/2017 12:38 2 Comments

Fork me on GitHub