Hi Chris,

I have not checked to see Gecko's/WebKit's behaviour here. I planned to 
track their WPT results once the patch lands upstream
and file bugs/make fixes as needed.
Let me know if we think this is worth verifying on other browsers 
beforehand and/or if I should reach out to those folks too!

I had assumed RUM providers wouldn't need to be notified but was hoping for 
feedback from this list on the subject 😄.
I doubt there would be problems because the empty string ought to already 
come up for them quite often when dealing with
cross-origin resources but I don't know if that's a good enough bar or 
not... It certainly wouldn't hurt to bring it up on the
WebPerf Working Group call tomorrow; I'll take an AI to poll the folks over 
there for their opinions as I know at least a few RUM
providers typically attend.

I'm hoping to get this landed this week (modulo feedback, of course) with 
no backporting/merging so would expect it to roll
out in M99.

The CL <http://crrev.com/c/3354335> extends the WPT of `nextHopProtocol 
<https://source.chromium.org/chromium/chromium/src/+/main:/third_party/blink/web_tests/external/wpt/resource-timing/nextHopProtocol-is-tao-protected.https.html>`
 
to cover the new behaviour. The test was previously written to only test 
fetching
iframe content so we didn't have coverage for this specific issue before 
😬. The way the spec is written, `nextHopProtocol`
should not have this special case so our updated WPT will exercise the 
spec'd behaviour more comprehensively.

Let me know if you have any other questions!

Cheers,
- Tom

On Wednesday, January 5, 2022 at 12:26:54 PM UTC-5 Chris Harrelson wrote:

> Hi Tom,
>
> Four questions:
> * Do Webkit and Gecko implement the same behavior already?
> * Will RUM providers need to be notified?
> * In what milestone will this be shipped?
> * I assume that the behavior you're implementing is already specified and 
> tested in WPT?
>
> On Wed, Jan 5, 2022 at 9:10 AM Tom McKee <[email protected]> wrote:
>
>> Howdy!
>>
>> I'm working to land a CL <http://crrev.com/c/3354335> to fix 
>> crbug.com/1284631 to better align our implementation of Resource Timing 
>> with the Resource Timing <https://www.w3.org/TR/resource-timing-2> and 
>> Fetch <https://fetch.spec.whatwg.org/> specs.
>>
>> The result of this change will affect all `non-document destination` 
>> fetches (e.g. fetches that aren't iframe contents). Such fetches used to 
>> always expose the, potentially sensitive, `nextHopProtocol` value. After 
>> this change, we will treat `nextHopProtocol` the same as any other 
>> sensitive information in Resource Timing; expose it to same-origin 
>> requesters or when the response's Timing-Allow-Origin header explicitly 
>> allows the requester's origin.
>>
>> This change is web observable as folks that have been reading the 
>> `nextHopProtocol` value could start to see the empty string instead of 
>> things like "http/1.1", "h2", etc. AFAICT, there's only three classes of 
>> users who could be impacted:
>>
>>    - *Well-meaning/typical content*: We don't expect this change to 
>>    break anything here because Resource Timing is a diagnostic API.
>>    - *RUM providers*: They'll see shifts in their data if/when they 
>>    collect `nextHopProtocol` but there are already many cases where they'd 
>>    need to handle the empty string already. 
>>    - *Shady folks:* Sniffing other origin's protocol values will no 
>>    longer be possible in this way which is probably what we'd have liked 
>> from 
>>    the start :)
>>
>> Let me know if folks think this change warrants sharing to a larger 
>> audience, though!
>>
>> Cheers,
>> - Tom
>>
>> P.S. crbug.com/1284631 is basically the same issue but for the 
>> `workerStart` attribute. A fix for that should be coming down the pipe 
>> shortly :)
>>
>> -- 
>> You received this message because you are subscribed to the Google Groups 
>> "blink-dev" group.
>> To unsubscribe from this group and stop receiving emails from it, send an 
>> email to [email protected].
>> To view this discussion on the web visit 
>> https://groups.google.com/a/chromium.org/d/msgid/blink-dev/CAD%2BxnEFZa1DjWf2Ap83tf9fRu6p4e2XtVMyA2kV0WeVXWYT3yA%40mail.gmail.com
>>  
>> <https://groups.google.com/a/chromium.org/d/msgid/blink-dev/CAD%2BxnEFZa1DjWf2Ap83tf9fRu6p4e2XtVMyA2kV0WeVXWYT3yA%40mail.gmail.com?utm_medium=email&utm_source=footer>
>> .
>>
>

-- 
You received this message because you are subscribed to the Google Groups 
"blink-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To view this discussion on the web visit 
https://groups.google.com/a/chromium.org/d/msgid/blink-dev/9ab29af4-e1d9-487b-af56-c97c73175869n%40chromium.org.

Reply via email to