mapbox-machine-styles to latest

[yhahn]

Updates to latest styles. This should be ready to go for next release.

cc @springmeyer @mtirwin

[springmeyer]

@yhahn - great, thanks for queuing this up. I've added to v0.3.3 milestone (#1491). What should the changelog entry be for this?

[yhahn]

Updated to mapbox-machine-styles 4.0.1

would be fine. It looks like appvey is failing with some encrypted credentials problems -- is this expected?

[springmeyer]

It looks like appvey is failing with some encrypted credentials problems -- is this expected?

Not expected. The weird thing is that appveyor is passing in your branch, just not in master:

https://ci.appveyor.com/project/Mapbox/mapbox-studio-classic/history

/cc @BergWerkGIS and I were speaking the other day about this oddity: why does appveyor run a branch build twice and how do they differ. I think @BergWerkGIS understood this but I don't recall what the answer was...

So, to recap I'm wondering if this is actually okay and the credentials problem will go away once merged.

[wilhelmberg]

why does appveyor run a branch build twice

I think two builds are run, because of a new commit to a PR triggers two events:

• push
• synchronizing PR

AppVeyor webhook settings:

You can see there are two SHAs (clickable, https://ci.appveyor.com/project/Mapbox/mapbox-studio-classic/history) for the commit:

and how do they differ

My hunch is, that sometimes, some tests take longer than expected (e.g. AppVeyor under heavy load) and the previous test has not yet finished, thus causing the current one to fail:

I've already experienced this before (not even on AppVeyor, but locally on Windows) and added some delay to make tests pass, e.g.:

        setTimeout(function(){
            t.end();
        }, 1000) 

mapbox-studio-classic/test/source.test.js

Lines 35 to 38 in fc368e6

	setTimeout(function(){
	console.log('after tm.config - ending test');
	t.end();
	}, 1000)