1220313 - Implement more proxied http- observers for Mochitest-plain

Status: RESOLVED WONTFIX

(Core :: DOM: Core & HTML, defect)

Description

[Andrew McCreight [:mccr8]]

nsObserverService::AddObserver() checks if the topic starts with "http-on" and rejects the observer. There's already a specialpowers-http-notify-request that proxies things through the parent, so we could hopefully implement some more of these to enable more tests.

The tests are:
  dom/base/test/test_bug1011748.html       http-on-examine-response
  dom/base/test/test_x-frame-options.html  http-on-examine-response
  dom/xml/test/test_bug392338.html         http-on-modify-request
  dom/security/test/csp/test_blocked_uri_in_reports.html   http-on-opening-request
  dom/security/test/csp/test_report.html   http-on-opening-request

Comment 1

[Andrew McCreight [:mccr8]]

test_x-frame-options.html only uses this observer for some debugging in bug 589457, for an issue which was fixed 4 and a half years ago, so that can just be fixed by removing it. These are all pretty separate so maybe I'll just split this bug up.

Comment 2

[Andrew McCreight [:mccr8]]

Boris, can I remove dom/xml/test/test_bug392338.html? Bug 392338 comment 8 says "The bug here is that the XMLDocument is set as the notification callbacks on the channel, and implements nsIInterfaceRequestor, and QIs to it, but does NOT expose it in classinfo.  Which means XPConnect throws when you try to get the callbacks from the channel, because it thinks it can't create the wrapper." which does not sound super relevant in the WebIDL era.

When I ran the test locally, it does the check that notification callbacks of the nsIChannel is in fact nsIInterfaceRequestor, but it isn't an instance of nsIClassInfo, so it doesn't do the rest of that check, which seems like it would be the interesting thing. As you would expect, as XMLDocument does not have anything in nsDOMClassInfo.cpp.

Comment 3

[Andrew McCreight [:mccr8]]

I'll just ask for review in bug 1220730.

Comment 4

[Andrew McCreight [:mccr8]]

I ended up filing separate bugs for these tests. The work they need to do in the parent observer is complex enough that using a chrome script makes more sense than trying to come up with a generic proxy to the child.