Iglistkit: Fix broken builds on travis-ci

[jessesquires]

Not sure what the deal is because travis is pretty much always broken af.

Looks like XCTestExpectations are timing out.

Maybe we increase the timeout?

[Sherlouk]

To be fair the tests weren't working when I ran them locally yesterday (same tests)

Not convinced it's 100% to the fault of Travis.

[Iron-Ham]

Same experience as @Sherlouk yesterday

[jessesquires]

lol. classic.

see, this is what happens when CI is totally unreliable.

[jessesquires]

also cc @rnystrom

we should probably at least increase those timeouts to like 20

[Sherlouk]

Is it worth giving something like Circle or Buddybuild a go? Had positive experiences with both of them in the past?

[Iron-Ham]

@Sherlouk I've used both. My experience re: Travis v Circle is that Circle has about a third of the total time in CI and I found it to be more reliable. I hope y'all don't use the built in travis-cron feature, otherwise you're going to have to spin up a herokuapp that does it (it's what I did).

BuddyBuild vs Travis: Similar to Circle in that it's much faster, but packed with features that are really more useful for a production app; not so much a framework. More expensive than Circle. Great for tracking down production crashes, _a la_ Fabric, but renders a video of the last few seconds before the crash and sends it your way.

[jessesquires]

we need free for OSS

[Sherlouk]

Circle & Buddybuild both have free packages for OSS projects

[rnystrom]

@jessesquires @Sherlouk there's a flaky test we have disabled internally that #657 fixes, but I don't think that has anything to do w/ the timeouts.

[jessesquires]

@rnystrom hm, yeah i'm getting a ton of failures and bullshit locally

especially on tvOS

[rnystrom]

@jessesquires which tests? The one I know about is

test_whenDeletingItemsTwice_withDataUpdatedTwice_thatAllUpdatesAppliedWithoutException

[jessesquires]

@rnystrom -- that one yes, but also see a ton of weirdness and/or flakiness

[tapanprakasht]

@rnystrom It looks like build issue is resolved. So I think this issue needs to be closed.