Backout for regressions with some major web services.

Approval Request Comment [Feature/regressing bug #]: 1097464 [User impact if declined]: "Adding channel description" of YouTube is broken. [Describe test coverage new/current, TreeHerder]: https://treeherder.mozilla.org/#/jobs?repo=try&revision=a398a6968f6b [Risks and why]: This is a big chunk of changes, needs time to get feedback and do tests. [String/UUID change made/needed]: r=mattwoodrow

Comment on attachment 8709480 [details] [diff] [review] Backout for Aurora 45 Let's do that. Thanks

Looks like Tomcat pushed this to Aurora today and it stuck. https://hg.mozilla.org/releases/mozilla-aurora/rev/64ec448f156d

OK, glad it stuck. We are moving 46 to aurora very soon. So this backout will have to happen all over again next week.

Backing this out on 46 will mean APZ gets pushed back, which means e10s gets pushed back. I strongly urge you to find another solution.

I did not realize this had anything to do with APZ or e10s. Then we should address all the regressions remaining quickly in aurora (where we will also be uncovering new regressions) with a plan to assess readiness to move to beta. If it isn't ready, then yes, we will have to push back e10s, APZ, and this work for 46. Maybe we don't care that it's a lot of regressions that are user visible; we are now trying to determine through telemetry whether those regressions are bad enough to drive users away, or whether shipping them is better than what we have now because of the overall issues of stability and sandboxing we will fix with e10s. In any case, the answer really shouldn't be, make it hard or impossible to back out the code as is the issue here.

(In reply to Liz Henry (:lizzard) (needinfo? me) from comment #6) > I did not realize this had anything to do with APZ or e10s. Then we should > address all the regressions remaining quickly in aurora (where we will also > be uncovering new regressions) with a plan to assess readiness to move to > beta. If it isn't ready, then yes, we will have to push back e10s, APZ, > and this work for 46. Maybe we don't care that it's a lot of regressions > that are user visible; we are now trying to determine through telemetry > whether those regressions are bad enough to drive users away, or whether > shipping them is better than what we have now because of the overall issues > of stability and sandboxing we will fix with e10s. > > In any case, the answer really shouldn't be, make it hard or impossible to > back out the code as is the issue here. Bug 1168263 is an APZ blocker, and it was one of the dependencies that got backed out here. Lets keep pushing to get the regressions fixed, and we can consider backing out later on in the Aurora cycle.

Could you please update the state of all of the bugs that were backed out, with comments pointing to relevant changesets? And is it correct that all of these bugs were also either backed out in 44 or never landed in 44?

Ok! I would do it in following days.

Ryan had updated the state of all of bugs. Thanks!