Fail aborted navigation

[sadym-chromium]

#799

---

Preview | Diff

[whimboo]

I would still like to see a debate if from a user perspective this is a acceptable solution. Comparing to our WebDriver classic implementation (where we do not fail) this would be a regression for those who want to switch to WebDriver BiDi. It's unclear how many pages will be affected and we know two major ones already.

Should the usage of location.href for redirecting a load really end-up as a failed navigation? Also in case of x.com I can see that they now use <meta http-equiv="refresh" content="0; url = ... for the redirect.

[OrKoN]

I believe from the user perspective the current behavior could be considered a bug if the user specifies waiting for the load event in the navigate call because that condition is not met on the success response. The abort errors gives a clearer understanding that the current navigation was not completed and the client needs to consider additional events. Alternatively, we could specify how waiting conditions could be fulfilled tolerating an aborted navigation but then it's not clear what the command should return since it's dealing with multiple navigations. I think adding the navigation committed wait condition (as requested in other issues) would allow not throwing an abort error.

[whimboo]

Just to add a reference to the issue Alex mentioned in his last comment about the navigation committed event: #788