Map style changes on OpenStreetMap.org | Hacker News Reader

Map style changes on OpenStreetMap.org

blog.openstreetmap.org

60 pointschippy23 comments

I like the constant updates and community behind OpenStreetMap. I'm thinking, would there be a particular reason why a developer would choose Google Maps instead of OpenStreetMap for a new application? I don't know if there are any major flaws

The main reasons that someone would still choose Google, as I (a long-term OSM developer) see it:

- Address search. OSM has lousy address coverage.

- One single provider, works out of the box. OSM is much more flexible in the long run, but you can't just plug in an "OpenStreetMap API" and expect it to do everything for you. (That said, Mapbox is going a long way towards providing this.)

- Satellite imagery. (Again, you can get this from a reseller like Mapbox.)

- StreetView.

- Public transport directions.

- Familiarity.

https://www.mapillary.com/ is working on a crowd sourced Creative Commons licensed street view data source.

Looks really cool. I am doing a project that relies heavily on Google Street View photos, and having an open dataset would be amazing. Is there any guarantee that Mapillary will always offer the user-contributed photos to the public for free download? And if they disappear, that all the photos will still be available?

How could OSM's address coverage be improved?

Go to http://www.openstreetmap.org/, find your street, click 'Edit', login or create an account, and make sure your local area is correctly classified. You can mark an area as a building and give it tags to say what the address and street number is.

I don't think Google mapped every street single address to a set of lat/long?? There must be mathematical formula to map a street address, knowing how the city allocated street numbers

Oh dear. Maybe in some places, which has a sensible and consistant addressing format. But there are plenty of places where there is no rhyme or reason to it.

Have a look at Germany in OSM. That's what crowdsourced maps can do.

Actually they do. This is why what you enter in CAPTCHAs appears to be house numbers - it is from Google Streetview.

They pull in public data. OpenAddresses is doing that and gives a sense of just how much is available:

http://openaddresses.io/

You also see Google talking about extracting building numbers from Streetview photos.

Same way any aspect of OSM is improved! By editing the map! Get out there and find some addresses and add them to OSM>

Here's some examples:

* Telnav Scout: a typical GPS/navigation app. I guess figure using OpenStreetMaps is better than their proprietary data, and they don't want to be competing with Google and depending on Google for data at the same time.

* Humanitarian Organizations (HOT): it turns out that you can get volunteers to trace maps from satellite imagery for remote places that have no other maps.

* Backcountry/mountaineering: a lot of places in the mountains are poorly mapped in Google Maps. OSM is a lot easier to edit, and supports a lot more features when it comes to describing terrain and whatnot.

Only one I know of is data availability in an area you need, or possibly better routing support (I haven't done anything with that in OSM before).

Check out http://project-osrm.org/ for routing.

Disclosure: I work for Mapbox, we offer a directions service based on OSRM, I'm paid to work on it daily.

As well as OSRM, there's also GraphHopper.

I don't understand this part:

> In the longer term there is the option to repopulate the database used by the Standard map style so that it has access to all OpenStreetMap data tags, not just a limited few.

How does a map 'style' repopulate the database? I get that not all tags can be displayed (e.g. many businesses have a website, but you don't display that unless you want to view the business's details) so why would a map need "access" to all data tags?

I have to be reading it wrong, but I don't see how.

There's the core OSM database (plain ol' Postgres), which contains all current OSM data plus all historical edit data. This is what you change if you click 'Edit' on OSM.

There's also the rendering database (PostGIS), which contains a subset of the current data, with freeform tags moved into Postgres columns, and with PostGIS geometries created from the raw OSM geodata. This is continually updated from the diffs of the core database. It's used by Mapnik (rendering code) to create the images you see on osm.org.

The proposed change is to add an hstore column to the rendering database, which will be populated with all the tags for each object from the core OSM database. This will require a complete reimport of the data.

Where would be a good place to start if I wanted to learn about adding additional data layers (in this case, US parcel ownership data) to OSM?

uMap is great for that:

https://wiki.openstreetmap.org/wiki/UMap

If you want to do more complicated things, try leaflet, http://leafletjs.com/

Can you explain what you mean? The OSM data base (which you can download), doesn't have "layers". It's all just one big topologically connected pile of objects.

You are free to make a local map with OSM data, and then, in your browser, or GIS software of choice, overlay anything else you want on it.

I parse that as 'There is a database. Which we use as source for the style/rendering process. It's a subset of our data, which means the style cannot even access everything we got. We might expand the subset/render from the real thing in the future'

Don't worry - it's an internal implementation thing about how the information trickles down to the intermediary database used for the rendering.

I think what they mean is that the previous Standard map style, which is produced by Mapnik, was rendered in such a way that the rendering process didn't have access to all of the possible OSM tags. But in the future they can use Mapnik to render more tags than they have in the past because Mapnik has improved in some way.

Reply on news.ycombinator.com