Wednesday, September 14, 2016

MapKit Compatibility Roadmap

We get asked this one a lot:  "Why can't you be more like MapKit?"  It's a good question.

MapKit, obviously.

Internally the toolkit won't ever be more like MapKit.  But we can pretend in a couple of ways:  Overlays and Interfaces.

MapKit Overlay

MapKit provides overlay options for everything from markers to tiled image sources.  Annotations are super flexible, but very slow beyond a handful.  And don't even think about animated tile sources.  Ouch.

For some data sources it's going to be easier to render in WhirlyGlobe-Maply and slap the results on top of MapKit's UIView.

It's not topical, I just like MapKit's stations.

That's the solution.  We'll track what MapKit is up to, at least in flat map mode, and overlay our own rendering to match.  Think weather apps.

MapKit Interfaces

MapKit has a nice collection of interfaces for adding annotations, great circles, and all sorts of other good stuff.  We'll try to be more compatible with those.

Portland in MapKit.  Why not?

This means methods that take MKMapPoint or CLLocationCoordinate2D objects.  It means init calls that are similar to the MapKit equivalents.  And yes, it means working in (shudder) degrees.

The goal here is a simpler conversion from a MapKit app to a WhirlyGlobe-Maply app.

UIView vs UIViewController

This one really bugs a handful of developers.  Our main interfaces are derived from UIViewController rather than UIView.

View Controllers are nice because they have life cycle events.  You have to wire all that up yourself in a UIView.  But it's possible and we'll do it because we know you want it.

Roadmap Not Actual Size

That's the plan and this all goes in the base toolkit so it'll be free.  But we need money to do it.  Always with the money.

As with all our plans, it'll get done eventually.  Pieces will go into proposals, clients will step forward, and some of it we'll just do ourselves.

Got feedback?  Let us know.

Wednesday, September 7, 2016

Android Tutorials

We have tutorials!  For Android!  Yay!

Screen shot of a web page?  I'm not proud.

A big thanks to Nicholas for putting these together.  And to José for doing the binary AAR examples.

Getting Set Up

The first few tutorials focus on setting up a really basic project and getting the WhirlyGlobe-Maply library included.  We now have three ways of doing this from least Android-y to most Android-y.

Yes, we're doing nightly builds for Android and we have the JCenter/Maven thing set up.  Once you've got the toolkit included, you can peruse the tutorials on actually doing stuff.

Enough With the Typing

There's nothing more to say.  Go check out the tutorials if you're doing Android development with WhirlyGlobe-Maply.

If you'd like more tutorials on iOS or Android, open up an Issue.  That's how we communicate.

We can thank the Support Customers for this since it came out of their budget.  More support contracts means a bigger budget.  It's all spent on stuff like this.

Tuesday, September 6, 2016

Mapzen Vector Tile Service

We're going to support Mapzen's Vector Tile Service.  Currently that means we read their vector tiles.  In the future we'll support their style sheets.

Courtesy Mapzen

Let's do a little background and then get to the details.

Vector What Now?

We've been slinging around vectors for years in WhirlyGlobe-Maply so vector tiles weren't a huge diversion.  We even had our own format for a while.

These days vector tiles means Mapbox Vector Tile format, but it implies two other things:
  • A curated version of OpenStreetMap suitable for web and mobile map display
  • One or more style sheets to describe the visual representation

You might call this a Vector Tile Service if you were so inclined.  Developers want to shove a map underneath the thing their app is actually doing.

It's a trivial, seamless process with an image tile service.  We're going to make it just as easy for a vector tile service.

Here is our roadmap to get there.

Mapzen Vector Tile Service

We can parse and display Mapzen's vector tiles just fine in either GeoJSON or Mapbox Vector Tile format.  But we use really simple style sheets.

Must be night or something

That looks kind of meh.  For a real map we need better.  There are a number of ways to do style sheets, but I'd like something that requires minimal upkeep.

Mapzen has a Javascript/C++/Java based renderer called Tangram.  It's got it's own style sheet format.  That format is.... very specific to Tangram. 

I've been on the fence about their style format.  Some discussion was in order.

Data Formats & Common Usage

Data formats have crazy parts.  Go look at PDF (better yet, don't).  You have to decide what's common usage and what parts you ignore, at least for a while.

The Mapzen folks were kind enough to host a meeting where we could explore this.  It looks like the common usage among their various style sheets is blunting the weirdness.  They've got professional cartographers using it, in addition to engineers, which is a big plus.

In short, when we see some deep_Tangram_weirdness we can substitute minor_WG-Maply_weirdness and it'll look pretty similar.  Good enough. 

There's Open and Then There's Open

The most attractive thing about the Mapzen Vector Tile Service is the terms of service.  You can use the data in any API and in a variety of ways.  As long as you're not completely rude.

Best of all, you can mix online and offline databases without violating the license.  That makes it perfect to slip into a WhirlyGlobe-Maply app and in our test apps.

That's not a knock on other services.  I get that the API and data are often tightly linked by a license.  That's how you make money.  And speaking of money...

Roadmap and Funding

The vector tile part works just fine in WhirlyGlobe-Maply now.  We still need to implement the Tangram style format.  We're just wrapping up Styled Layer Descriptor support so it shouldn't be all that difficult.

It does require money.

That's the roadmap.  If you're interested in any part of it, speak up.  I get asked about this a lot, so it'll get paid for eventually.