Add criticalCHRestart

[arichiv]

Website can indicate that a particular Client Hint is critical to the page by including it in a Critical-CH HTTP response header. Doing so will trigger a connection restart if the hint listed in the Critical-CH HTTP response header could be (but wasn't) included in the HTTP request initially sent. We should indicate this has happened in Navigation Timing so that developers can know if a restart occurred and impacted timing.

We can pull this from the navigable but I'm not quite sure if this is the right way to thread it.

Closes #177

---

Preview | Diff

[noamr]

Is this specific to navigations? Subresources don't support this behavior?

[arichiv]

Yes that's correct. Just top-level navigations too (not iframes).

[noamr]

Yes that's correct. Just top-level navigations too (not iframes).

Great!
Do we know the timing of that restart? Maybe it should be a timestamp rather than a boolean? Isn't there some point in time where the UA decided to re-send the navigation or so?

[arichiv]

Done @noamr

[noamr]

OK I'm missing two things here. without them this feels a bit hypothetical:

1. where do we set this new value
2. Web platform tests that show how it's used

[arichiv]

If this looks good otherwise, I believe that:

1. Is resolved by modifying https://wicg.github.io/client-hints-infrastructure/ to set it and
2. Is resolved after Chrome implements and tests.
   Sounds like this is paused until after those happen.

[noamr]

It's true this isn't reflected in other specs, including the DNT spec itself: "Ideally, the value of Navigator.doNotTrack ought to reflect the current set of user-granted exceptions in effect when the attribute is read." https://www.w3.org/TR/tracking-dnt/#expressing

I think it's worth requiring locked consistency within a given navigation for all subresources and subframes as that seems the most correct/consistent approach and this spec is still being incubated, but I'd also be fine with unifying HTTP/JavaScript approaches the other direction and always using the 'live' value of the preference if that's what you're saying.

Not sure what you're responding to. Perhaps a different issue?

[arichiv]

Sorry, yes. Just updated the comment

[noamr]

If this looks good otherwise, I believe that:

1. Is resolved by modifying https://wicg.github.io/client-hints-infrastructure/ to set it and
2. Is resolved after Chrome implements and tests.
   Sounds like this is paused until after those happen.

Yea, would like to see the whole picture. It's difficult to review addition of a new attribute without this context.

[arichiv]

Awesome, will check back in a couple weeks or so when that's done.

[arichiv]

The WPTs just went in: https://chromium-review.googlesource.com/c/chromium/src/+/4538244
Is this the right way to do the patching to set the time in the spec: WICG/client-hints-infrastructure#153

[arichiv]

@noamr any feedback on WICG/client-hints-infrastructure#153?

[arichiv]

The other PR is in, can this be committed @noamr?