Fixes WaitForNavigation race condition in #125

[ap0]

When multiple redirects happen very quickly, sometimes the event handler will not be removed before the next page is loaded, so it will call the handler twice and panic because it's attempting to close a previously-closed channel.

Closes #125

[codecov]

Codecov Report

Merging #281 into master will increase coverage by <.1%.
The diff coverage is n/a.

@@           Coverage Diff            @@
##           master    #281     +/-   ##
========================================
+ Coverage      56%   56.1%   +<.1%     
========================================
  Files         189     189             
  Lines        5777    5777             
========================================
+ Hits         3238    3240      +2     
  Misses       2193    2193             
+ Partials      346     344      -2

[ziflex]

Good use case. I wonder how stable it is in this case. If many redirects occur, it handles only first one, it misses the rest which might also lead to some unpredictable behavior.

[ap0]

Yeah, I don't know enough about how this works to say. Right now, though, it crashes whenever it encounters this. Would it be better to somehow have the event broker map events to specific pages when it loads instead?

[ziflex]

What if we create a debounce function to handle that?
Something like this:
https://gist.github.com/jonstaryuk/133ced9882aa00502fa7d1db31cb5253

[ap0]

I like that solution, but what about the case where it's in the middle of a redirect when the debounce function finally calls the callback (i.e. one of the intermediate pages is slow but will still redirect)?

The big thing for me here is to just guard against calling close() on a channel twice, which I don't think that would guarantee, but would make the chances of it happening less likely.

[ziflex]

Good point. I think in this case it may just fallback to this initial solution with sync.Once.

[ziflex]

Ok, I will merge it. But we need to improve it in the future.