1249172 - Make nsHttpAuthManager service e10s compliant

Status: RESOLVED WORKSFORME

(Core :: Networking: HTTP, defect, P3)

Description

[Kyle Machulis [:qdot] [:kmachulis] (INACTIVE)]

In bug 1237834, we were getting oranges in e10s tests. This is due to nsHttpAuthManager existing in both the parent and content processes, and information between the two getting desynced. We should only have an nsHttpAuthManager living in the parent process, and any content reference to the process should access that instance.

Comment 1

[Brad Lassey [:blassey] (use needinfo?)]

Should this block the roll out of e10s?

Comment 2

[Kyle Machulis [:qdot] [:kmachulis] (INACTIVE)]

I don't think so? The only test it affected was the one that's listed as a blocker on this bug, and that's fixed. All other tests that use it are either marked as skip on e10s (e.g. password manager tests), or don't break because they use anonymous credentials. mccr8, you got any opinion on this?

Comment 3

[Andrew McCreight [:mccr8]]

I don't really know anything about this, sorry.

jdm, it looks like you were the last person to make substantative changes to this class, back in 2012. Is this just not being used for much if we only notice in a few tests that it is broken in e10s?

Comment 4

[Josh Matthews [:jdm]]

The only non-test usage of nsIHttpAuthManager (or the concrete class, nsHttpAuthManager) is in nsNPAPIPlugin.cpp. I have no idea how important retrieving HTTP authentication credentials is for Flash, but this feels like something that isn't a high priority in the list of e10s things.

Comment 5

[Firefox Bug Husbandry Bot]

Bulk change to priority: https://bugzilla.mozilla.org/show_bug.cgi?id=1399258

Comment 6

[Firefox Bug Husbandry Bot]

Bulk change to priority: https://bugzilla.mozilla.org/show_bug.cgi?id=1399258

Comment 7

[:Gijs (he/him)]

e10s has shipped and plugins are gone -> wfm.