Handle Promise rejection in guards (fix #2833)

[oleksiikhr]

#2833

[posva]

Hi, thanks for the PR!
However, we don't use babel to transpile the lib, so this won't work (it is used with tests). Could you find a version without adding the babel plugin dependency?
Can you also add the tests for the others navigation guards, so they are all covered? beforeEnter, beforeRouteEnter, beforeRouteUpdate
Also, could you please add back the template for the PR? It helps us to organise things

Thanks

[oleksiikhr]

Very interesting, solved without a library :)

Also, could you please add back the template for the PR? It helps us to organise things

What is a template?

[posva]

Hey! I realized you closed the issue. Did something happen? Maybe it was closed by accident? Eduardo San Martin Morote

…

On 2 Jul 2019, at 22:09, Alexey Khrushch ***@***.***> wrote: Closed #2838. — You are receiving this because you commented. Reply to this email directly, view it on GitHub, or mute the thread.

[oleksiikhr]

I studied this code, found 2 places (hook) where Promise can be accepted (beforeEach, afterEach):

https://github.com/vuejs/vue-router/pull/2838/files#diff-feafce5e816936212eccd6808c5421a6R123
https://github.com/vuejs/vue-router/pull/2838/files#diff-feafce5e816936212eccd6808c5421a6R185

And if in the first case the execution of the code, the sequence itself does not change, in the second case it is not so (R185).

If we take into account that everything happens through callbacks, then in the function confirmTransition, I also need to transfer to onComplete the callback also the this._abort function. In the abstract.js, base.js files, receive and transmit the updateRoute function.

In general, I do not like it, I understand that most likely I need to wrap parts of the code in Promise, do consistent error handling and not break the original logic.

Since this PR has errors in logic, I decided to close it, so that suddenly due to inactivity, they would not begin to accept it :)

[posva]

Thanks for the explanation. I wanted to make sure it wasn't an accident :)
With current codebase it's a bit harder to implement indeed