Intervention header mismatch
Reported by
ve...@ecwid.com,
Oct 25
|
|||||||||
Issue descriptionSteps to reproduce the problem: Perform page request on slow connection to achieve https://www.chromestatus.com/features/4775088607985664 Browser sends >Intervention <https://www.chromestatus.com/features/6072546726248448>; level="warning" header for my "script.js" and >Intervention <https://www.chromestatus.com/features/4775088607985664 >; level="warning" header for my favicon What is the expected behavior? What went wrong? It should be vice versa: 6072546726248448 for favicon 4775088607985664 for script.js Did this work before? N/A Chrome version: 69.0.3497.100 Channel: stable OS Version: 8.1.0 Flash Version:
,
Oct 26
,
Oct 26
verde@ -- Thanks for reporting this issue. Could you please help us by providing the detailed screencast reproducing the issue. This will help us in understanding the issue better. Also please share the device details, OS version on which the issue is reproduced. You can also verify by updating your device to latest Stable #70.0.3538.64 and let us know your observations. Thanks in advance!
,
Oct 30
Interesting. I looked at the code and the right header should be added for the right optimization, but as I understand it, the headers should only be added for the frame request. Adding dougarnett@ to look into it.
,
Oct 30
Maybe my understanding is wrong, and every request in the frame gets the same header added.
,
Oct 30
I think the LoFi intervention header (6072546726248448) should be added to all subresources (since in general it is not clear ahead of time whether the resource will end up being an image or not). So that first case might be expected - if LoFi was triggered for that page load. But I believe we only want the NoScript header (4775088607985664) on the main frame request, so I can fix that.
,
Oct 31
My understanding was that for a given navigation, all subresources should get the same entry for the intervention header. I am not sure why we are adding different entries for different resources.
,
Oct 31
cc-ing bengr The original intent for NoScript was to add the header just on the main frame (I double checked also in go/cacao-client-design). We could discuss further about Selective Resources Blocking (since originally we wanted to use Reporting API and it is not mentioned in the design doc) but it certainly couldn't be on the blocked subresource so main frame makes sense with ability for the Developer to return cache-control:no-transform header on main frame response.
,
Oct 31
Discussed offline. Lets also add the intervention header on browser initiated navigations. That would go somewhere here: https://cs.chromium.org/chromium/src/content/browser/frame_host/navigation_request.cc?rcl=a8234d9a57aceea5014e8865da19cfea3f2764f0&l=183
,
Nov 5
,
Nov 6
Discussed offline with bengr@ and others. We want to deprecated the intervention header in favor of using the new Reporting API so deprioritizing making changes to existing intervention header behavior. Plan to replace with Reporting API in M-73.
,
Nov 20
Issue 887657 tracks the work of using Reporting API to report previews interventions. Can we mark this as WontFix?
,
Nov 20
|
|||||||||
►
Sign in to add a comment |
|||||||||
Comment 1 by eroman@chromium.org
, Oct 25