[BUGFIX beta] fixes transitionTo failing when called in beforeModel hook

[mattmarcum]

This is a fix for 11563.

This bug is caused when a queryParamOnly transitionTo is called in the beforeModel hook.

There's a couple of problems here. One is that route.finalizeQueryParams expects the route's controller to already be setup, but it isn't until the route.setup hook is called. The problem is that route.setup is not called until after the beforeModel hook is run.

The solution I came up with is to extract out the relevant portions of the controller setup from route.setup and then check during route.finalizeQueryParams to see if the controller has been added, and if not, add it to the route.

There was another problem with updatePaths on the router not exiting out quick enough. @raytiley has a similar fix in #13273 but in my case the infos param was not even an array.

[mattmarcum]

All the tests pass locally I promise! I think the build is failing because of a deprecated package in npm.

[rwjblue]

I think its fixed (we cleared the Travis cache), going to restart the build...

[rwjblue]

So we don't want to stashParamNames if there are no handlerInfos? This is to ensure that #setup has been called first, right? If so, is there a better way to check this (or add a comment that explains why)?

[raytiley]

This looks good to me with the caveat that my eyes start to bleed when staring at at the router code for too long.

This will be great cause this issue is what I was blocked on for #13273. Once this is in I'll finish up some of the other tests and then query params not bug people so much for a while :)

[rwjblue]

I think that this augments #13273 pretty well. I had a couple of minor nitpicks that I'd like to see addressed, but other than that I'm pretty happy with this. Thank you for taking the time to work on it!

[mattmarcum]

@rwjblue fixed those. I used transition.queryParamsOnly when deciding whether to call stashParamNames. The queryParamsOnly flag gets set when the no-op transition gets created (hence the reason the handlerInfos are empty). I think it makes it more clear.

[rwjblue]

Awesome, thank you! Will merge once Travis agrees....

[mattmarcum]

Oops, looks like that last change broke some other tests...I'm going to revert back and just add a comment instead.

[homu]

☔ The latest upstream changes (presumably b125ff5) made this pull request unmergeable. Please resolve the merge conflicts.

[homu]

☔ The latest upstream changes (presumably ceb2b15) made this pull request unmergeable. Please resolve the merge conflicts.

[mattmarcum]

After looking at this code again, I'm thinking that a better solution would need to touch the router.js code too. I'm going to close this pr and think about a better solution. Any one else, feel free to try and fix this.