Extend browser.createUserContext with unhandledPromptBehavior parameter

[sadym-chromium]

Addressing #789

Analogue of acceptInsecureCerts, add unhandledPromptBehavior parameter to browser.createUserContext.

---

Preview | Diff

[burg]

I prefer the separate command that configures the session behavior, optionally for a user context. At least compared to a capability, it would make it easier to share implementations in WebKit. The current unhandledPromptBehavior capability is already quite complicated and I'm not sure why it has to be a capability anyway.

[sadym-chromium]

I prefer the separate command that configures the session behavior, optionally for a user context. At least compared to a capability, it would make it easier to share implementations in WebKit. The current unhandledPromptBehavior capability is already quite complicated and I'm not sure why it has to be a capability anyway.

I'm not sure how having a separate command would ease the implementation, as we need to specify the behavior per user context.

[css-meeting-bot]

The Browser Testing and Tools Working Group just discussed Extend `browser.createUserContext` with `unhandledPromptBehavior` parameter.

The full IRC log of that discussion

<tidoust> Topic: Extend `browser.createUserContext` with `unhandledPromptBehavior` parameter
<tidoust> github: https://github.com//pull/896
<tidoust> sadym: After the first extension to createUserContext, I have a couple of new proposals which are PRs.
<tidoust> ... For unhandledPromptBehavior, two approaches are possible. Time to discuss and make a decision!
<jgraham> q+
<jimevans> ack next
<tidoust> jgraham: I agree that we shouldn't do more with capabilities here.
<tidoust> ... When we create a user context, at that point, you can specify unhandledPromptBehavior for the entire context, which is more fine-grained than what we have today, but less fine-grained than per browsing context.
<tidoust> ... It's not clear to me what the objection to this PR is. It seems fine. Later on, if we want to do it per browsing context, we can. For now, per user context seems a nice start to address use cases we know about without adding too much complexity.
<tidoust> jimevans: Seems that we can get to consensus on this one!
<tidoust> [no objection heard]