Closed
Bug 997808
Opened 11 years ago
Closed 10 years ago
[e10s] HTTP redirects don't redirect when HTTPS Everywhere addon is enabled
Categories
(Core :: Networking: HTTP, defect)
Core
Networking: HTTP
Tracking
()
RESOLVED
FIXED
People
(Reporter: cpeterson, Unassigned)
References
()
Details
(Keywords: dogfood)
STR:
1. In an e10s window, load http://mozilla.org/
RESULT:
Instead of redirecting to http://www.mozilla.org/en-US/, Firefox displays the intermediate redirect page:
Moved Permanently
The document has moved here.
I think this is a recent regression.
Updated•11 years ago
|
Component: Networking → Networking: HTTP
Comment 1•11 years ago
|
||
Chris, this is working for me in an e10s window on OSX 10.9 with Nightlies 4/15, 4/16 and today's 4/17. Can you still reproduce it? If so, any other info you can provide?
Flags: needinfo?(cpeterson)
| Reporter | ||
Comment 2•11 years ago
|
||
This only happens when I have the HTTPS Everywhere addon enabled.
Blocks: e10s-addons
Flags: needinfo?(cpeterson)
| Reporter | ||
Comment 3•11 years ago
|
||
This is not a regression after all. I just had the HTTPS Everywhere addon disabled until recently.
No longer blocks: core-e10s
Keywords: regression,
regressionwindow-wanted
OS: Mac OS X → All
Hardware: x86 → All
Summary: [e10s] HTTP redirects don't redirect → [e10s] HTTP redirects don't redirect when HTTPS Everywhere addon is enabled
Comment 4•11 years ago
|
||
Ah, I see. Still might be worth looking into.
| Reporter | ||
Updated•11 years ago
|
No longer blocks: e10s-addons
The redirects should work once ContentPolicy is not registered anymore in chrome.manifest (which it shouldn't be since we stopped using ContentPolicy a while ago). I fixed this in the stable git branch: https://github.com/efforg/https-everywhere/tree/4.0
You can run ./makexpi.sh to build an xpi to try it out, or wait til the next release.
| Reporter | ||
Comment 6•11 years ago
|
||
Thanks, Yan!
Comment 8•11 years ago
|
||
Still seems to be broken on the https-everywhere 4.0 branch, rev d0638d8e39cd. Firefox from current mozilla-central (gecko-dev git rev 77fe533a054f).
Updated•11 years ago
|
QA Whiteboard: qawanted
Comment 9•11 years ago
|
||
Jacob Hoffman-Andrews is working on HTTPS Everywhere for EFF/Tor, but I'm not sure if he's the new permanent go-to guy. cc:ed; can you help here?
Flags: needinfo?(mozilla.20.jsha)
| Reporter | ||
Comment 10•11 years ago
|
||
* Jacob: I installed https-everywhere-4.0development.17.xpi and, with e10s enabled, I see the following error message in Firefox's Browser Console when trying to load "cnn.com":
> NS_ERROR_UNEXPECTED: Component returned failure code: 0x8000ffff (NS_ERROR_UNEXPECTED) [nsILoadContext.associatedWindow] https-everywhere.js:425
* Ally: this looks like another nsILoadContext.associatedWindow problem. Are you planning to shim associatedWindow?
Flags: needinfo?(ally)
Comment 11•11 years ago
|
||
afaik, objects aren't eligible for shims, but I think its an excellent candidate for a CPOW.
Flags: needinfo?(ally)
Comment 12•11 years ago
|
||
Try building from https://github.com/diracdeltas/https-everywhere/tree/electrolysis (started working on e10s compatibility a long time ago).
Comment 13•11 years ago
|
||
(In reply to yan from comment #12)
> Try building from
> https://github.com/diracdeltas/https-everywhere/tree/electrolysis (started
> working on e10s compatibility a long time ago).
Redirects in e10s tabs work for me in that version.
| Reporter | ||
Comment 14•11 years ago
|
||
Adding "dogfood" keyword so this bug shows up on the e10s wiki's list of known issues:
https://wiki.mozilla.org/Electrolysis#Known_Issues
Keywords: dogfood
Comment 15•11 years ago
|
||
After today's update https-everywhere still broken with e10s (36a1). Tried fresh profile and 5.0 development version of https-everywhere.
Comment 16•10 years ago
|
||
I have a similar problem, when there is an HTTPS redirect, the page is simply blank, nothing displayed. If I manually put https:// before the URL all works fine.
Comment 17•10 years ago
|
||
We've landed Yan's electrolysis branch on master but haven't released it yet. I was doing some pre-release testing and found that non-e10s HTTPS Everywhere now has some regressions, related to getting null from channel.loadGroup. I'm at a bit of a dead-end. If any Mozilla folks would like to hop into the discussion on https://github.com/EFForg/https-everywhere/pull/526, or join me in #necko on irc.mozilla.org, I'd greatly appreciate the help.
| Reporter | ||
Comment 18•10 years ago
|
||
Allison, can you take a look at the HTTP Everywhere error? It's odd that channel.loadGroup would work with e10s, but return null in non-e10s builds.
Flags: needinfo?(ally)
Comment 19•10 years ago
|
||
Blake, I know nothing about channels, but I agree that something is fishy if it works with e10s but returns null for single process.
I'm hoping you could shed some light on the issue
Flags: needinfo?(ally) → needinfo?(mrbkap)
Comment 20•10 years ago
|
||
With the new 5.0developement.1 build of HTTPS Everywhere, this issue looks now works for me. I tested with the mozilla.org example and received the local version from the redirection.
| Reporter | ||
Comment 21•10 years ago
|
||
Thanks for testing, alreiten. 5.0development.1 build works for me, too. I'd like to keep this bug open until HTTPS Everywhere's stable release is e10s compatible (i.e. updates from 4.0 to 5.0).
Updated•10 years ago
|
Flags: needinfo?(mrbkap)
Comment 23•10 years ago
|
||
As of HTTPS Everywhere 5.0.0, the stable release is e10s compatible. Please report any compatibility bugs to our GitHub issues page: https://github.com/EFForg/https-everywhere/issues
| Reporter | ||
Comment 24•10 years ago
|
||
Thanks, Jacob!
Status: NEW → RESOLVED
Closed: 10 years ago
Flags: needinfo?(mozilla.20.jsha)
Resolution: --- → FIXED
You need to log in
before you can comment on or make changes to this bug.
Description
•