Open Bug 1755064 Opened 2 years ago Updated 2 years ago

BroadcastChannel Storage Key: make webmessaging/broadcastchannel/cross-partition.https.tentative.html pass

Categories

(Core :: DOM: postMessage, enhancement)

enhancement

Tracking

()

UNCONFIRMED

People

(Reporter: awillia, Unassigned)

Details

Steps to reproduce:

Per https://github.com/whatwg/html/pull/7567, the spec is being updated to mention that BroadcastChannel should be partitioned by storage key instead of origin. Right now this is a no-op, since the storage spec only includes origin in the storage key, but this is expected to change soon per https://github.com/whatwg/storage/issues/127

Opening this bug report for tracking.

Related: https://bugzilla.mozilla.org/show_bug.cgi?id=1753569

Thank you for filing this! It is always very nice to have a direct link to spec changes!

I understand our code to currently be using our storage key, but that it's likely the policy that we're running the https://searchfox.org/mozilla-central/source/testing/web-platform/tests/webmessaging/broadcastchannel/cross-partition.https.tentative.html WPT test under is not doing the additional keying that would result in this test passing. Presumably we would want to add a pref to https://searchfox.org/mozilla-central/source/testing/web-platform/meta/webmessaging/broadcastchannel/cross-partition.https.tentative.html.ini if the test itself doesn't have a way of telling our test runner this wants to run partitioned... :pbz, do you have any idea about this situation like how to best enable TCP in WPTs? Thanks!

Flags: needinfo?(pbz)
Summary: Partition BroadcastChannel by storage key → BroadcastChannel Storage Key: make webmessaging/broadcastchannel/cross-partition.https.tentative.html pass

Redirect to Ben who has worked on enabling dFPI for our tests.

Flags: needinfo?(pbz) → needinfo?(bvandersloot)

Hi all.

Broadcast Channels are consistently a bother with TCP, and it is good to have the spec language as to why. It makes sense that the WPT is only origin keyed given that.

Given the current spec language I think that you are right, we should add a pref to disable TCP for that test. Alternatively we could declare ourselves as failing that test.

To disable TCP for the test, it is a one line diff to insert prefs: [network.cookie.cookieBehavior:4] below line 1 of https://searchfox.org/mozilla-central/source/testing/web-platform/tests/webmessaging/broadcastchannel/cross-partition.https.tentative.html .

Flags: needinfo?(bvandersloot)
You need to log in before you can comment on or make changes to this bug.