> https://bugzilla.mozilla.org/show_bug.cgi?id=1072980 > > was filed 6 weeks ago, and diagnosed as relating to force rtl (but Firefox's > fault, because we're passing CPOWs to functions that should never have them) > within a few days, after which nothing happened.
ForceRTL is not a commonly installed add-on, which has reduced this bug's priority. > I duped the bug I filed to an earlier bug, courtesy of mak for finding > those: > > https://bugzilla.mozilla.org/show_bug.cgi?id=1026521 > > That was filed 4.5 months ago. This was believed to be mitigated (by disabling e10s when hardware acceleration was disabled), but thanks to your testing we discovered a bug in that mitigation: https://bugzilla.mozilla.org/show_bug.cgi?id=1095559 This is exactly the kind of testing we are hoping to get. We are making Nightly "rougher" to use (particularly this week as we shake out the major issues), but as Dave mentioned, we are doing this because it will help us ship e10s. Gavin On Fri, Nov 7, 2014 at 8:33 AM, Gijs Kruitbosch <gijskruitbo...@gmail.com> wrote: > On 07/11/2014 16:26, Dave Townsend wrote: > >> wrote: >> >>> Where can we read about how this decision was made? On my profiles with >>> Force RTL, e10s perma-crashes ( https://bugzilla.mozilla.org/ >>> show_bug.cgi?id=1072980 ), and in most other cases, I only ever see >>> spinners for remote content (ie can't actually browse the web - doesn't >>> matter too much for me because I normally use beta). >>> >>> I recognize that my experience is not universal, but I hope it makes >>> sense >>> that I'm... surprised... that we're turning this on for everyone. >>> >> >> As you might expect the decision was made based on us believing that we >> had >> had enough testing to have a good idea that the remaining bugs weren't bad >> enough to severely affect usage but wanted the additional testing that >> turning on in Nightly would give us. >> >> While most of the e10s team have been running with e10s enabled for a >> while >> at a certain point (as for any feature) we have to increase the number of >> testers to learn about bugs that we don't see in our configurations and >> usage patterns. We started that with prompting users to opt-in to e10s >> around a month ago and since then around 25% of Nightly users have been >> running with e10s enabled. As we hoped this saw a surge in the number of >> bugs filed and helped us understand the remaining issues more. Many many >> bugs have been fixed since then and as of Tuesday nothing remaining looked >> bad enough to block us turning this on more globally. >> >> The crash sounds like a late breaking change that should be fixed >> imminently, that's just the peril of nightly. > > > https://bugzilla.mozilla.org/show_bug.cgi?id=1072980 > > was filed 6 weeks ago, and diagnosed as relating to force rtl (but Firefox's > fault, because we're passing CPOWs to functions that should never have them) > within a few days, after which nothing happened. > >> Not being able to browse the >> >> web at all is an entirely new one that we hadn't seen in our testing to >> date. While it is unfortunate that this has broken the browser for you >> (and >> undoubtedly others) this is exactly the outcome we want for enabling a >> feature on nightly. More testing to reveal the bugs that the developers >> don't see themselves. Hopefully we'll get the most serious ones fixed >> quickly. If we can't then we get to talk about whether turning e10s off >> again is the right solution. >> >> I'm interested to know if you had opted in to e10s at all in the past and >> seen this issue with the permanent spinner? That might help us narrow down >> when it started to be a problem. > > > I duped the bug I filed to an earlier bug, courtesy of mak for finding > those: > > https://bugzilla.mozilla.org/show_bug.cgi?id=1026521 > > That was filed 4.5 months ago. > > Both of these were marked blocking-e10s+ (although the top one got reset to > ? just now - it's not clear to me why that happened). > > ~ Gijs > > _______________________________________________ > dev-platform mailing list > dev-platform@lists.mozilla.org > https://lists.mozilla.org/listinfo/dev-platform _______________________________________________ dev-platform mailing list dev-platform@lists.mozilla.org https://lists.mozilla.org/listinfo/dev-platform
